Introduction

Only a few epidemiology studies have been able to link dietary intake of polyunsaturated fatty acids (PUFAs) to breast cancer risk (Wolk et al, 1998a; Velie et al, 2000; Michels et al, 2001; Wirfält et al, 2002), but many animal, pharmacological and clinical studies suggest crucial roles for PUFAs in breast cancer development (Okuyama et al, 1997; Stoll 1998; Howe et al, 2001). A meta-analysis of animal experiments, with mutual adjustment for different fats, indicated a strong tumour enhancing effect of 6 fatty acids, a weaker enhancing effect of saturated fat, while 3 fatty acids had a small, nonsignificant protective effect (Fay et al, 1997). Foods rich in saturated fats have in case–control studies been positively related to breast cancer (Boyd et al, 1993). Case–control studies among Mediterranean populations have observed negative relations with olive oil (Trichopoulou et al, 1995; WCRF/AICR (JD Potter), 1997). Although these relations have been interpreted as protection by monounsaturated fat (MUFA), confounding by other components in olive oil or by plant foods is plausible. As food sources of fatty acids commonly overlap, results from different epidemiological studies are often contradictory (Byers & Gieseker, 1997). However, a few cohort studies have examined different types of fat in mutually adjusted models (Wolk et al, 1998a; Holmes et al, 1999; Velie et al, 2000; Smith-Warner et al, 2001). The Nurses Health Study (Holmes et al, 1999) and the international pooling project (Smith-Warner et al, 2001) did not observe any relations for the different types of fat. Velie et al (2000), however, observed a positive breast cancer relation for unsaturated fat in postmenopausal women without previous benign breast disease. Wolk et al (1998a) observed that among Swedish women, MUFA was inversely and PUFA positively related to breast cancer. Also, women of the latter cohort with high consumptions of linoleic acid had lower breast cancer risk if intakes of ascorbic acid were high (Michels et al, 2001).

Contradictory findings in epidemiology studies could depend on measurement errors in dietary assessment, causing both differential and nondifferential misclassification of intakes and weakened associations (Wong et al, 1999; Kipnis et al, 2001). Nutrition epidemiologists and statisticians therefore strongly advocate the use of biomarkers in order to corroborate intake measures and to reflect internal exposures (Byers & Gieseker, 1997; Kohlmeier, 1997). Although good biomarkers of dietary intake are rare, both feeding experiments and methodological studies indicate that the fatty acid composition in biological tissues like serum lipoproteins, erythrocytes and adipose tissue quite readily reflect the type of fat in the diet (Zock et al, 1997). Tissue concentrations of exogenous PUFAs (for instance, linoleic acid and fatty acids in fish oils) clearly reflect dietary intakes (Zock et al, 1997). Some saturated fatty acids (SFAs) (eg 15:0 and 17:0 in dairy fat) are also suitable as dietary biomarkers (Wolk et al, 1998b; Smedman et al, 1999), but fatty acids synthesised endogenously (eg oleic acid) are not. Experimental studies have indicated that erythrocytes are sensitive indicators of dietary change in polyunsaturated fat (Zock et al, 1997). The fatty acid composition of erythrocyte membranes depends both on the fatty acids present at the time of erythropoiesis and on the current plasma concentration (Hunter, 1998). Thus the fatty acid composition of erythrocyte membranes reflects dietary fat intakes over a few to several weeks.

A review of biomarker studies on fatty acids and breast cancer indicates that tissue concentrations of 6 fatty acids were generally positively related to breast cancer risk, but protective effects of 3 fatty acids were less consistent (Kohlmeier, 1997). A European multicentre study examining ratios of 3 to 6 fatty acids in adipose tissue in relation to breast cancer risk (Simonsen et al, 1998) found consistent inverse associations of the fatty acid ratios with breast cancer across most study centres (Kohlmeier, 1997). Two breast cancer studies that examined fatty acids in phospholipids (Chajès et al, 1999) and erythrocytes (Pala et al, 2001) suggest complex relations between diet, fatty acid and hormone metabolism, and breast cancer.

Obesity is an established risk factor for postmenopausal breast cancer. However, central obesity and weight gain may be more specific markers of the metabolic consequences of obesity than general obesity (Kaaks, 1996). Energy-dense diets, low in fibre but high in fat, alcohol and added sucrose promote positive energy balance and contribute to the development of obesity (Bray & Popkin, 1998). Also, some researchers argue that saturated fat is more important in obesity development than unsaturated fat (Wang et al, 2002). In contrast, a few studies have suggested that SFAs originating from milk are negatively related to metabolic risk factors like insulin resistance (Smedman et al, 1999).

After menopause, much of the circulating oestrogen is derived from aromatisation of androgen in peripheral adipose tissue, resulting in higher circulating oestrogen concentrations in obese women compared to lean (Stoll, 1994). Stress (eg anxiety, depression, alcohol, smoking, overfeeding) may in some individuals lead to increased concentrations of cortisol, which stimulates visceral deposition of fat, accompanied by insulin resistance in muscle tissues and high concentrations of insulin-like growth factors (IGFs) (Björntorp, 1999). It is believed that both sex hormones (especially free oestradiol) and IGFs may act (possibly in synergism) as breast cancer promoters by stimulating the mitotic division of initiated breast cancer cells (Kaaks, 1996; WCRF/AICR (JD Potter), 1997; Yu & Berkel, 1999). A few epidemiological studies support the involvement of IGFs in breast cancer development (Yu & Berkel, 1999). Although breast cancer and type II diabetes share a number of risk factors (Sinagra et al, 2002), the relations between the two appear complex in epidemiological studies (Sellers et al, 2003). However, a report from the Nurses Health Study suggests that type II diabetes may be associated with an increased risk of breast cancer, especially in postmenopausal women and in those with oestrogen receptor-positive tumours (Michels et al, 2003).

Similar to other westernised countries, the breast cancer rates vary for different areas of Sweden (National Board of Health and Welfare, 2003). Higher incidence rates are observed in the south (ie 309.64 cases per 100 000 in the province of Skåne in the year 2000) compared to the country as a whole (ie 288.32 cases per 100 000). Malmö is the third largest city of Sweden and a major commercial centre in the Southern region. The Malmö Diet Cancer (MDC) is a population-based prospective cohort with access to dietary data of high concurrent validity and stored blood samples (Berglund et al, 1993). The MDC offers a unique setting for dietary biomarker studies. A previous analysis project suggested that high intakes of 6 fatty acids were associated with increased risk of breast cancer in postmenopausal women (Wirfält et al, 2002). The current project uses the same study design and has three aims. (1) to examine the relations between the dietary intake of fatty acids and the fatty acid composition of erythrocyte membranes; (2) to examine the relations between indexes of general and central obesity and the fatty acid composition of erythrocyte membranes; and (3) to estimate the risk of breast cancer associated with the fatty acid composition of erythrocyte membranes.

Methods

The MDC study conducted baseline examinations from March 1991 to October 1996 (Berglund et al, 1993). The project presented in this paper is the second paper from a nested case–control study examining the relation between diet and postmenopausal breast cancer. The previous project examined the relation between different types of dietary fats and breast cancer (Wirfält et al, 2002). A third paper will examine food sources of fat and breast cancer.

Population: the Malmö Diet and Cancer cohort

The source population was first defined as all persons living in the City of Malmö and born between 1926 and 1945. In May 1995, the cohort was extended to include all women born between 1923 and 1950, and all men born between 1923 and 1945. With this extension, the source population consists of 74 138 individuals. Individuals of the cohort (men, n=11 063; women, n=17 035) either joined the study spontaneously or after receiving a mail-invitation. Limited Swedish language skills were the major exclusion criterion. The proportions of individuals with higher education and white-collar workers in higher positions are slightly greater, and the proportions of current smokers and individuals of non-Swedish origin are slightly smaller in the MDC cohort compared to the background population (Manjer et al, 2001). The ethical committee at Lund University reviewed and approved the MDC study protocol.

Ascertainment of cancer cases and selection of controls

This study used an age-based definition of menopause (Morabia & Flandre, 1992), and subsequently a total of 12 803 women aged 50 y and more were eligible for the study. Prevalent breast cancer cases (n=515) were excluded. A total of 249 incident breast cancer cases occurred from the time of the baseline examinations until end of follow-up (31 December 1999). Cancer cases were ascertained by record linkage with the Swedish Cancer Registry and the Southern Swedish Regional Cancer Registry. Women without breast cancer (n=12 039) at the time of study entry and during follow-up were eligible controls. Three controls (alive at the time of breast cancer diagnosis) were matched to each case on year and month of birth, and year and month of the first visit to the study centre. Individuals with other prevalent cancer at study entry were excluded (ie, 12 cases and 39 controls), except those with cancer cervix in situ and nonmalignant skin cancer. Due to the matched design, an additional 35 controls were removed from analysis because the corresponding cases were excluded. No case–control set was excluded due to prevalent cancer among the controls only. Thus, this project examined 237 case–control sets (237 incident breast cancer cases and 673 matched controls).

Data collection

Participants came to the study centre at two occasions. At the first visit, trained project staff provided groups of participants with detailed instructions about the data collection procedure, distributed the study questionnaires, and conducted anthropometrical measurements. At this visit, participants also donated nonfasting blood samples. At a second visit (approximately 2 weeks later), trained dietary interviewers conducted the diet history interview and checked the correctness of questionnaires completed at home.

Dietary data

Diet history data were obtained by combining a 7-day menu book ('current' diet information about cooked meals and cold beverages, recorded by participants at home) and a questionnaire ('usual' diet information about foods consumed regularly and not eaten during cooked meals). The reference period of the questionnaire was the preceding year. A total of 17 dietary interviewers conducted interviews focusing on food preparation and portion sizes reported in the menu book, and checking questionnaires completed at home. An interactive computer software program (Kostsvar, AIVO AB, Stockholm), which facilitated standardised data entry and homogeneity across interviewers, was used. The MDC food and nutrient database, specifically developed for the MDC study, originates from PC KOST version 2/93 of the Swedish National Food Administration.

The reproducibility and concurrent validity of the diet methodology was examined with 18 days of weighed food records collected during 1 y as the reference (Elmståhl et al, 1996; Riboli et al, 1997). That study included 241 Malmö residents (126 men and 115 women) in the age range 50–69 y. The method overestimated total energy by 18% at group level compared to the reference method. The validation correlations were generally higher than those found in comparable dietary methods in other populations (Willett, 1998). Energy-adjusted correlations in women were: for total fat 0.69; saturated fat 0.68; monounsaturated 0.66; polyunsaturated fat 0.64; and for the ratio of PUFAs and SFAs 0.74. Correlations were: for the 6 fatty acids: linoleic acid (octadecadienoic) 0.68 and arachidonic (eicosatetraenoic) acid 0.44, and for the 3 fatty acids: linolenic (octadecatrienoic) acid 0.58, timnodonic (eicosapentaenoic) acid 0.38, clupanodonic (docosapentaenoic) acid 0.40 and cervonic (docosahexaenoic) acid 0.27.

Blood samples and biochemical analysis of fatty acids

Blood samples were processed and separated within 1 h of drawing, as described previously (Pero et al, 1993). Erythrocyte samples (3.0 ml) were washed with saline (10 ml) at 600 g for 10 min, and stored at -80°C before analysis. The quality of the samples has been described previously (Pero et al, 1998). The fatty acid composition of erythrocyte membranes was analysed at the Unit for Clinical Nutrition Research, Department of Public Health and Caring Sciences, Uppsala University, Sweden. The first step to determine the fatty acid composition was to add 2 ml of methanol to 0.5 ml of erythrocytes. Next, chloroform (5 ml), containing 0.005% butylated hydroxytoluene as an antioxidant, was added, followed by 7.5 ml of 0.2 mol/l sodium dihydrogen phosphate (NaH₂PO₄). After thorough mixing, the extract was left at +4^oC for 1–4 days. The chloroform phase was evaporated to dryness under nitrogen and, after addition of 2 ml 5% H₂SO₄ in methanol, transmethylated at 60^oC overnight. All work involving chloroform was performed in cupboards under ventilated hoods in accordance with the safety regulations for handling of these types of chemicals. After adding 1.5 ml distilled water and mixing thoroughly, the methyl esters were extracted into 3 ml of petroleum ether (b.p. 40–60°C) containing 0.005% butylated hydroxytoluene. The phases were separated by centrifugation at 1500 g for 10 min. The methyl esters were redissolved in 0.3 ml Uvasol grade hexane. The fatty acid methyl esters were separated by gas–liquid chromatography on a 25-m WOCT (wall-coated open tubular) glass capillary column coated with SLP OV-351 (Quadrex, New, CT, USA), and with helium as carrier gas. A Hewlett-Packard system (Avondale, PA, USA), consisting of model GLC 5890, integrator 3396 and auto-sampler 7671 A, was used. The temperature was programmed to 100–210°C. The fatty acids were identified, by comparing the retention time of each peak with the Nu Check Prep standards (Elysian, MN, USA) of fatty acid methyl esters. The purity of these standards is greater than 99%.

Fatty acid variables

The original dietary variables identify the absolute amounts consumed of each fatty acid. When analysing these variables (with three decimal places), it appeared that three fatty acids (ie 20:5, 22:5 and 22:6) had zero as the lowest intake value. These values were replaced with the lowest observable value (ie 0.001 g for five observations of 20:5, 0.002 g for two observations of 22:5 and 0.002 g for one observation of 22:6). Then the per cent contribution of each fatty acid to total intake of fatty acids was computed.

The fatty acid composition of erythrocyte membranes was expressed as the per cent of each fatty acid to the total sum of fatty acids. The original variables of 14:0, 18:3,3, 20:5,3 and 22:5,3 had observations with 'values below detectable level'. These values were set to the lowest observed value of each variable (ie 0.22 for one observation with 14:0, 0.12 for 10 observations with 18:3,3, 0.50 for one observation with 20:5,3, and 1.06 for two observations with 22:5,3). Most variables had near-normal distributions. One exception was the 15:0 variable, which had one very diverging (outlier) observation. This observation was recorded as a 'missing value'.

Indexes of desaturase enzyme activity

Ratios of fatty acids in biological samples are used as indicators of the activity of desaturase enzymes (Pala et al, 2001). As the ratios of fatty acids in biological samples change in response to dietary changes, these ratios can also be regarded as dietary markers (Vessby et al, 2002). The delta9-desaturase enzyme converts palmitic and stearic acids to palmitoleic and oleic acids (ie 16:0 to 16:1,7 and 18 to 18:1,9). The delta6-desaturase converts 18:2,6 to 18:3,6, and delta5-desaturase converts 20:3,6 to 20:4,6 (ie arachidonic acid) (Mayes, 1990). The following ratios of fatty acids from the erythrocyte membranes were computed: 16:1,7 to 16:0; and 18:1 to 18:0 (ie to reflect 'delta9 activity'); 20:3,6 to 18:2,6 (ie to reflect 'delta6 activity', assuming that the elongation of 18:3,6 to 20:3,6 is not a rate-limiting step in the metabolism of 18:2,6); and 20:4,6 to 20:3,6 (ie to reflect 'delta5 activity').

Sums and ratios of fatty acids

The sums of 3 and 6 PUFAs, PUFAs and SFAs, and the ratios of these variables (ie the 3–6 and the P–S ratios) were computed both for erythrocyte membrane and dietary fatty acids.

Other variables

Information on age was obtained through the person-identification number (in Sweden, each person is assigned a 10-digit number at birth; six digits indicate the date of birth and one identifies sex/gender). Information on socioeconomic status, education, smoking, alcohol and physical activity habits, past food habits change, and medical and health history was obtained through a standardised questionnaire. Detailed description of these variables is available elsewhere (Wirfält et al, 2001).

Height (m), weight (kg) and waist circumference (cm) were obtained through direct measurements without shoes and in light indoor clothing. Body mass index (BMI) was computed as the ratio between weight (kg) and height raised to the second power (m²). Central adiposity was defined as having a waist circumference of 80 cm or more. A continuous variable age-at-menarche was constructed from the self-reported year of menarche and the year of birth (obtained from the personal identification number). Two categorical variables were constructed from the self-reported year of birth of each child: parity (no children, one child, two or more children) and age-at-birth-of-first-child (no children, below 24 y, 24–29.9 y, 30 y and above). The 'no children, categories of these variables include women who did not give any information concerning 'the year of birth of each child'. Information on the current use of hormone therapy was obtained from two sources: the diet history menu book, where the use was recorded during 7 consecutive days; and the sociodemographic and lifestyle questionnaire (Merlo et al, 2000). The information was combined into a dichotomous variable (yes; no).

Statistical analysis

All analyses used the SPSS statistical computer package (version 9.0; SPSS Inc., Chicago, IL, USA). All tests of statistical significance were two-sided. The partial correlations between the dietary proportions of fatty acids (DFA) and the proportions of fatty acids of erythrocyte membranes (EFA) were computed among the controls (n=673), controlling for total energy and age at baseline. Also, Spearman's rank order correlations were computed between EFA and obesity indexes among the controls.

The distribution of body composition indexes and macronutrients was described in cases and controls, as well as the mean proportions and percentile distributions of EFA and DFA. The risks of postmenopausal breast cancer associated with all EFA, with the EFA ratios of P–S and 3–6, and with the EFA indexes of desaturase activity were estimated using conditional logistic regression analysis. Each variable was examined separately. Analysis was repeated with multivariate models controlling for height, waist circumference, BMI, hormone therapy use, age-at-birth-of-first-child and alcohol habits. A previous project has concluded that these variables showed significant or borderline associations with breast cancer risk, and are thus potentially confounders of diet–breast cancer relations in the Malmö Diet Cancer study (Wirfält et al, 2002).

Finally, the relations between quintiles of desaturase activity indexes and breast cancer risk were examined separately for those with and without central adiposity in multivariate models. Five exposure categories were created based on the quintile ranking of controls. Cases were assigned to the categories depending on the enzyme index values.

Results

As can be seen from Table 1, cases were heavier and had wider waist circumferences compared to controls. Energy intakes appear similar, but the proportion of carbohydrates was larger among cases compared to controls, accompanied by a tendency of lower fat proportion among cases.

Table 1 - Characteristics of cases and controls.

Full table

Table 2 shows the descriptive statistics of dietary and erythrocyte membrane fatty acids in cases and controls. No differences are observed between cases and controls, except higher dietary intake of linoleic acid among cases, as shown in a previous study (Wirfält et al, 2002). The EFAs are very similar in cases and controls, and so are the indexes of desaturase activity and the P–S- and 3–6 ratios. Although slightly different in rank order, the proportions of three fatty acids—palmitic, oleic and linoleic acids—rank the highest both in DFA and EFA. However, arachidonic acid and the three 22-carbon fatty acids contribute a larger proportion of EFA than of DFA.

Table 2 - Descriptive statistics of dietary and erythrocyte fatty acids in cases (n=237) and controls (n=673) (values are per cent (%)).

Full table

The partial correlations between EFA and DFA in controls are shown in Table 3. Positive, comparatively low-grade, but highly significant correlations are seen between several fatty acids. The two dietary 'fish' fatty acids (ie 20:5 and 22:6) are clearly correlated with the corresponding EFA. Significant positive correlations are also seen between DFA 20:4 and EFA 22:6; and between DFA 14:0 and 16:0, and EFA 22:5. Similarly, the erythrocyte membrane 'milk' fatty acids (ie 14:0, 15:0 and 17:0) are correlated with DFA 14:0 and 16:0. In addition, the dietary linoleic acid is positively correlated with the erythrocyte membrane linoleic acid.

Table 3 - Partial correlations between erythrocyte fatty acids (EFA) and dietary fatty acids (DFA)^b among controls (n=673).

Full table

Table 4 shows the rank order correlations between EFA and obesity indexes in controls. Negative and significant relations are seen between the 'milk fatty acids' and the obesity indexes, especially for those indicating central obesity. Also, positive and significant relations are seen between the obesity indexes and the 16:1 and 20:3 fatty acids. Three of the four indexes of desaturase activity are significantly related to the obesity indexes.

Table 4 - Spearman's rank order correlations between erythrocyte fatty acids (EFA), indexes of desaturase enzyme activity, and obesity indexes, among controls (n=673).

Full table

The risk of breast cancer associated with EFA (as continuous variables) was estimated both with and without adjustment for confounders (Table 5). None of the examined fatty acid variables were associated with increased risk for breast cancer in the unadjusted multivariate models. However, borderline relations were observed for two of the desaturase indexes (ie 20:3/18:2 and 20:4/20:3) when adjusting for established risk factors and potential confounders.

Table 5 - Breast cancer risk associated with erythrocyte fatty acids in postmenopausal women (n=910) of the Malmö Diet and Cancer cohort.

Full table

When examining quintile categories of the desaturase indexes in adjusted models, separately for individuals with and without central obesity, borderline relations were only visible in the fifth quintiles among women without central obesity. The risk estimates were stronger, but did not reach significance (ie for 20:3/18:2, RR=0.50, 95% CI 0.23–1.07; and for 20:4/20:3, RR=2.07, 95% CI 0.97–4.40).

Discussion

This study observed significant positive associations between dietary fatty acids and erythrocyte membrane fatty acids from fish and milk, and for linoleic acid. The findings suggest that indexes of obesity (especially central obesity) are negatively related to milk fat. Positive associations were also seen between obesity indexes and activity of the delta9 and delta6 desaturase enzymes. However, the study did not observe any associations between the proportions of fatty acids in erythrocyte membranes and breast cancer risk.

Fatty acids in erythrocyte membranes reflect the dietary intakes over a period of some weeks. One could argue that this is a reasonable period for a dietary biomarker because diet memories decay with time, and 'generic' memories are mostly recalled when people are asked to report about their usual diet (Smith, 1991). Researchers have argued that erythrocytes are not very sensitive to dietary intakes (Pala et al, 2001), and that the fatty acid composition of erythrocyte membranes is under tight control. The lack of differences in EFAs between cases and controls, and the very small standard deviations (Table 2) observed in the current study of a cross-sectional population sample supports this argument. On the other hand, experimental feeding studies have found that erythrocyte membranes are specifically sensitive to PUFAs when intake change is monitored in the same individuals (Zock et al, 1997). In this study, positive correlations were observed between dietary and erythrocyte membrane fatty acids for long-chain polyunsaturated fish fatty acids, saturated milk fatty acids and for linoleic acid.

The correlation coefficients between dietary and erythrocyte membrane fatty acids were in this study lower compared with what has been observed in other biomarker studies (Bates et al, 1997; Hunter, 1998). A possible explanation is that this study used a comparatively large population sample with a variety of intakes, physiological states and tissue concentrations. Other studies may have used samples from specific population groups with more uniform correspondence between dietary fatty acids and tissues concentrations. It is, however, difficult to compare results from studies that assess fatty acid composition in different tissues. The observed concentrations of fatty acids are also dependent on the endogenous production of fatty acids, and the different roles of fatty acids in transportation and physiologic functions (Hunter, 1998).

Other Swedish studies have previously observed negative associations between cardio vascular risk factors and the SFAs specifically found in milk fat (Smedman et al, 1999). Similarly, in this study, all four of the examined obesity indexes were negatively related to 15:0 and 17:0. However, the correlation coefficients for waist circumference and waist–hip ratio (WHR) were higher than for BMI and body fatness. The causal background explaining this relation is unclear. Two food pattern studies from the MDC cohort indicate that the 3–6 ratio is higher among individuals selecting food patterns where milk fat provide large proportions of energy intake (Wirfält et al, 20002001). Also, the risk of insulin resistance is lower among women with this type of food pattern (Wirfält et al, 2001). A mixture of butter and rapeseed oil (ie Bregott^® margarine) is the major source of milk fat in this population. Thus a plausible explanation of the high 3–6 ratio (of diets high in milk fat in this population) is the covariation with rapeseed oil.

Breast cancer risk

One could argue that the dietary intakes reflected by the fatty acid composition of erythrocyte membranes cover too short a period to reflect accurately the latency period for breast cancer. Dietary exposures resulting in breast cancer are likely to last for many years (WCRF/AICR (JD Potter), 1997). In addition, the observed exposure may have little to do with the dietary exposures directly related to the development of breast cancer, because of dietary changes over time. Adipose tissue, which more passively reflects fat intakes over longer time periods, may therefore have been a better choice (Bates et al, 1997; Hunter, 1998). A European multicentre study examined fatty acids in adipose tissue samples and found consistent inverse associations of the 3–6 fatty acid ratios with breast cancer risk across most study centres (Kohlmeier, 1997). One could also argue that erythrocyte membranes may simply be the wrong tissue, because the specific processes that lead to cancer could only be studied in breast tissue. The preferred tissue as a biomarker of dietary fatty acids in relation to breast cancer risk would be breast adipose tissue. However, such tissue samples are very difficult to obtain. One case–control study that examined fatty acids from breast adipose tissue in breast cancer cases compared to controls with benign breast disease found a protective effect of 3 fatty acids (Klein et al, 2000).

Two studies examining fatty acids in relation to breast cancer risk, a Swedish study of serum-phospholipids (Chajès et al, 1999) and an Italian study of erythrocyte membranes (Pala et al, 2001), confirm that tissue concentrations of fatty acids are not simple reflections of diet, but the result of a multitude of metabolic activities. Both studies suggested that low activity of the delta-9-desaturase enzyme was associated with low breast cancer risk. The study by Pala et al reported high levels of MUFAs (ie the denominator of saturation index) in the erythrocyte membranes of breast cancer cases, resulting in an inverse relation between the saturation index and breast cancer risk. The Swedish study found high levels of stearic acid (ie the numerator of saturation index) in serum-phospholipids of the controls, also resulting in an inverse breast cancer relation with the saturation index. Other studies have shown that the activity of the delta9-desaturase enzyme is decreased by dietary PUFAs and fasting, but increased by dietary SFAs and carbohydrates, and by insulin, oestrogen and testosterone (Pala et al, 2001). The current study found no significant breast cancer associations with any fatty acid or ratio of fatty acids. However, the positive relations with 16:1/16 and 20:3/18:2 ratios and obesity indexes suggest that obesity is positively related to higher activity of the delta9- and delta6-desaturase enzymes. The negative relations with the 20:4/20:3 suggest that obesity is related to low activity of the delta5-desaturase enzyme. Insulin-resistant states are often characterized by plasma fatty acid patterns with high proportions of palmitic acid (16:0) and low proportions of linoleic acid (18:26), which indicates high activity of the delta9- and delta6-desaturases, and low activity of the delta5-desaturase (Vessby et al, 2002). The increased membrane saturation is a consequence of a high intake of dietary SFAs (Vessby et al, 2002). Thus the fatty acid patterns associated with abdominal obesity in this study are compatible with these previous observations. The relations between enzyme activity and breast cancer risk are, however, not clear, although the borderline significant observations in women without central obesity might indicate that the activities of the delta6- and the delta5-desaturase enzymes are differently related to breast cancer risk.

In conclusion, similar to other studies, this study observed significant positive associations between dietary fatty acids and fatty acids of erythrocyte membranes from fish and milk, and between dietary and erythrocyte membrane linoleic acid. Similar to other Swedish studies, biomarkers of milk fat showed negative relations with central obesity. The study suggests positive relations between indexes of obesity and desaturase enzyme activity, indicative of insulin-resistant states, but could not clearly support previous observations that the activity of the desaturase enzymes, or that -6 fatty acids, influence the development of breast cancer.

References

1. Bates CJ, Thurnham DI & Bingham SA (1997): Biochemical markers of nutrient intake. In Design Concepts in Nutritional Epidemiology, eds BM Margetts & M Nelson, pp 170–240. Oxford, New York: Oxford University Press.
2. Berglund G, Elmståhl S, Janzon L & Larsson SA (1993): The Malmö Diet and Cancer Study: design and feasibility. J. Intern. Med. 233, 45–51. | PubMed | ISI | ChemPort |
3. Björntorp P (1999): Obesity and diabetes mellitus. In Ellenberg & Rifkin's Diabetes Mellitus, eds D Porte & SR Sherwin Jr, pp 552–564. Stamford, CT: Appleton & Lange.
4. Boyd NF, Martin M, Noffel GA, Lockwood GA & Trichler DL (1993): A meta-analysis of studies of dietary fat and breast cancer risk. Br. J. Cancer 68, 627–636. | PubMed | ISI | ChemPort |
5. Bray GA & Popkin BM (1998): Dietary fat intake does affect obesity. Am. J. Clin. Nutr. 68, 1157–1173. | PubMed | ISI | ChemPort |
6. Byers T & Gieseker K (1997): Issues in the design and interpretation of studies of fatty acids and cancer in humans. Am. J. Clin. Nutr. 66(Suppl), 1541S–1547S. | PubMed |
7. Chajès V, Hultén K, Van Kappel AL, Winkvist A, Kaaks R, Hallmans G, Lenner P & Riboli E (1999): Fatty-acid composition in serum phospholipids and risk of breast cancer: an incident case–control study in Sweden. Int. J. Cancer 83, 585–590. | Article | PubMed | ISI | ChemPort |
8. Elmståhl S, Gullberg B, Riboli E, Saracci R & Lindgärde F (1996): The Malmö food study: the reproducibility of a novel diet history method and an extensive food frequency questionnaire. Eur. J. Clin. Nutr. 50, 134–142. | PubMed | ISI | ChemPort |
9. Fay MP, Freedman LS, Clifford CK & Midthune DN (1997): Effect of different types and amounts of fat on the development of mammary tumors in rodents: a review. Cancer Res. 57, 3979–3988. | PubMed | ISI | ChemPort |
10. Holmes MD, Hunter DJ, Colditz GA, Stampfer MJ, Hankinson SE, Speizer FE, Rosner B & Willett WC (1999): Association of dietary intake of fat and fatty acids with risk of breast cancer. JAMA 281, 914–920. | Article | PubMed | ISI | ChemPort |
11. Howe L, Subbaramaiah K, Brown A & Dannenberg A (2001): Cyclooxygenase-2: a target for the prevention and treatment of breast cancer. EndocrRelat. Cancer 8, 97–114.
12. Hunter D (1998): Biochemical indicators of dietary intake. In Nutritional Epidemiology, eds W Willett, pp. 174–243. Oxford, New York: Oxford University Press.
13. Kaaks R (1996): Nutrition, hormones and breast cancer: is insulin the missing link? Cancer Causes Control 7, 605–625. | Article | PubMed | ISI | ChemPort |
14. Kipnis V, Midthune D, Freedman LS, Bingham S, Schatzkin A, Subar A & Caroll RJ (2001): Empirical evidence of correlated biases in dietary assessment instruments and its implications. Am. J. Epidemiol. 153, 394–403. | Article | PubMed | ChemPort |
15. Klein V, Chajés V, Germain E, Schulgren G, Pinault M, Malvy D, Lefrancq T, Fignon A, Le Floch O, Lhuillery C & Bougnoux P (2000): Low alpha-linolenic acid content of adipose breast tissue is associated with an increased risk of breast cancer. Eur. J. Cancer 36, 335–340. | Article | PubMed |
16. Kohlmeier L (1997): Biomarkers of fatty acid exposure and breast cancer risk. Am. J. Clin. Nutr. 66(Suppl), 1548S–1556S. | PubMed |
17. Manjer J, Carlsson S, Elmståhl S, Gullberg B, Janzon L, Lindström M, Mattisson I & Berglund G (2001): The Malmö Diet and Cancer Study: representativity, cancer incidence and mortality in participants and non-participants. Eur. J. Cancer Prev. 10, 489–499. | Article | PubMed | ISI | ChemPort |
18. Mayes PA (1990): Metabolism of unsaturated fatty acids and eicosanoids. In Harper's Biochemistry, eds RK Murray, DK Granner, PA Mayes & VW Rodwell, pp 218–225. Norwalk, CT/San Mateo, CA: Appleton & Lange.
19. Merlo J, Berglund G, Wirfält E, Gullberg B, Hedblad B, Manjer J, Hovelius B, Janzon L, Hanson BS & Östergren PO (2000): Self-administred questionnaire is a simple and valid method compared to a personal diary for assessment of current exposure to hormone therapy—an analysis of 16 060 women. Am. J. Epidemiol. 152, 788–792. | Article | PubMed | ChemPort |
20. Michels KB, Holmberg L, Bergkvist L, Ljung H, Bruce A & Wolk A (2001): Dietary antioxidant vitamins, retinol, and breast cancer incidence in a cohort of Swedish women. Int. J. Cancer 91, 563–567. | Article | PubMed |
21. Michels KB, Solomon CG, Hu FB, Rosner BA, Hankinson SE, Colditz GA & Manson JE (2003): Type 2 diabetes and subsequent incidence of breast cancer in the Nurses' Health Study. Diabetes Care 26, 1752–1758. | PubMed |
22. Morabia A & Flandre P (1992): Misclassification bias related to definition of menopausal status in case–control studies of breast cancer. Int. J. Epidemiol. 21, 222–228. | PubMed | ISI | ChemPort |
23. National Board of Health and Welfare (2003): The Public's Health in numbers Cancer 1970–2000. [Folkhälsan I siffror Cancer 1970–2000.] Centre for Epidemiology, National Board of Health and Welfare, Stockholm Sweden. 4-30-2003. Ref Type: Electronic Citation.
24. Okuyama H, Kobayashi T & Watanabe S (1997): Dietary fatty acids–the N-6/N-3 balance and chronic elderly diseases. Excess linileic acid and relative N-3 deficiency syndrome seen in Japan. Progr. Lipid Res. 35, 409–457. | Article |
25. Pala V, Krogh V, Muti P, Chajès V, Riboli E, Micheli A, Saadatian M, Sieri S & Berrino F (2001): Erythrocyte membrane fatty acids and subsequent breast cancer: a prospective Italian study. J. Natl. Cancer Inst. 93, 1088–1095. | Article | PubMed | ChemPort |
26. Pero RW, Olsson A, Berglund G, Janzon L, Larsson SA & Elmståhl S (1993): The Malmö Biological bank. J. Intern. Med. 233, 63–67. | PubMed |
27. Pero RW, Olsson A, Bryngelsson C, Carlsson S, Janzon L, Berglund G & Elmståhl S (1998): Quality control program for storage of biologically banked blood specimens in the Malmö Diet and Cancer Study. Cancer Epidemiol. Biomarkers Prev. 7, 803–808. | PubMed |
28. Riboli E, Elmståhl S, Saracci R, Gullberg B & Lindgärde F (1997): The Malmö Food Study: validity of two dietary assessment methods for measuring nutrient intake. Int. J. Epidemiol. 26, S161–S173. | Article | PubMed |
29. Sellers TA, Anderson KE, Olsen JE & Folsom AR (2003): Family histories of diabetes mellitus and breast cancer and incidence of postmenopausal breast cancer. Epidemiology 9, 102–105.
30. Simonsen N, van' t Veer P, Strain JJ, Martin-Moreno JM, Huttunen JK, Fernández-Crehuet Navajas J, Martin BC, Thamm M, Kardinaal AFM, Kok FJ & Kohlmeier L (1998): Adipose tissue omega-3 and omega-6 fatty acid content and breast cancer in the EURAMIC study. Am. J. Epidemiol. 147, 342–352. | PubMed | ISI | ChemPort |
31. Sinagra D, Amato C, Scarpilta AM, Brigandi M, Amato M, Saura G, Latteri MA & Caimi G (2002): Metabolics syndrome and breast cancer risk. Eur. Rev. Med. Pharmacol. Sci. 6, 55–59. | PubMed |
32. Smedman A, Gustafsson I-B, Berglund L & Vessby B (1999): Pentadecanoic acid in serum as a marker for intake of milk fat: relations between intake of milk fat and metabolic risk factors. Am. J. Clin. Nutr. 69, 22–29. | PubMed | ISI | ChemPort |
33. Smith-Warner S, Spiegelman D, Adami H-O, Beeson WL, Van der Brandt PA, Folson AR, Raser GE, Freudenheim JL, Goldbohm RA, Graham S, Kushi LH, Miller AB, Rohan TE, Speizer FE, Toniolo P, Willett WC, Wolk A, Zeleniuch-Jacquotte A & Hunter D (2001): Types of dietary fat and bresat cancer: a pooled analysis of cohort studies. Int. J. Cancer 92, 767–774. | Article | PubMed | ISI | ChemPort |
34. Smith AF (1991): Cognitive processess in long-term dietary recall. Vital Health Stat. 6, 25–27.
35. Stoll BA (1994): Breast cancer: the obesity connection. Br. J. Cancer 69, 799–801. | PubMed | ChemPort |
36. Stoll B (1998): Breast cancer and the Western diet: role of fatty acids and antioxidant vitamins. Eur. J. Cancer 34, 1852–1856. | Article | PubMed |
37. Trichopoulou A, Katsouyanni K, Stuver S, Tzala L, Gnardellis C, Rimm E & Trichopoulos D (1995): Consumption of olive oil and specific food groups in relation to breast cancer risk in Greece. J. Natl. Cancer Inst. 87, 110–116. | PubMed | ChemPort |
38. Velie E, Kulldorff M, Schairer C, Block G, Albanes D & Schatzkin A (2000): Dietary fat, fat subtypes, and breast cancer in postmenopausal women: a prospective cohort study. J. Natl. Cancer Inst. 92, 833–839. | Article | PubMed | ChemPort |
39. Vessby B, Gustafsson I-B, Tengblad S, Boberg M & Andersson A (2002): Desaturation and elongation of fatty acids and insulin action. Ann. NY Acad. Sci. 967, 183–195. | PubMed | ChemPort |
40. Wang H, Storlien LH & Huang X-F (2002): Effects of dietary fat types on body fatness, leptin, and ARC leptin receptor, NPY, and AgRP mRNA expression. Am. J. Physiol. Endocrinol. Metab. 282, E1352–E1359. | PubMed | ISI | ChemPort |
41. WCRF/AICR (JD Potter) (1997): Food, Nutrition and the Prevention of Cancer: A Global Perspective. Washington, DC: World Cancer Research Fund/American Insitute for Cancer Research.
42. Willett W (1998): Nutritional Epidemiology, 2nd Edition, pp 101–147 Monographs in Epidemiology and Biostatistics, Vol 30. New York, Oxford: Oxford University Press.
43. Wirfält E, Mattisson I, Gullberg B & Berglund G (2000): Food patterns defined by cluster analysis and their utility as dietary exposure variables: a report from the Malmö Diet and Cancer Study. Public Health Nutr. 3, 159–173. | PubMed |
44. Wirfält E, Hedblad B, Gullberg B, Mattisson I, Andrén C, Rosander U, Janzon L & Berglund G (2001): Food patterns and components of the metabolic syndrome in men and women. A cross-sectional study within the Malmö Diet and Cancer cohort. Am. J. Epidemiol. 154, 1150–1159. | Article | PubMed | ChemPort |
45. Wirfält E, Mattisson I, Gullberg B, Johansson U, Olsson H & Berglund G (2002): Polyunsaturated fatty acids are associated with increased breast cancer risk in post-menopausal women of the Malmö Diet and Cancer cohort (Sweden). Cancer Causes Control 13, 883–893. | Article | PubMed |
46. Wolk A, Bergström R, Hunter D, Willett W, Ljung H, Holmberg L, Bergkvist L, Bruce Å & Adami H-O (1998a): A prospective study of association of monounsaturated fat and other types of fat with risk of breast cancer. Arch. Intern. Med. 158, 41–45. | Article | PubMed | ISI | ChemPort |
47. Wolk A, Vessby B, Ljung H & Barrefors P (1998b): Evaluation of a biological marker of dairy fat intake. Am. J. Clin. Nutr. 68, 291–295. | PubMed | ISI | ChemPort |
48. Wong MY, Day NE & Wareham NJ (1999): Measurement error in epidemiology: the design of validation studies II: bivariate situation. Stat. Med. 18, 2831–2845. | Article | PubMed |
49. Yu H & Berkel J (1999): Do insulin-like growth factors mediate the effect of alcohol on breast cancer risk? Med. Hypotheses 52, 491–496. | Article | PubMed |
50. Zock PL, Mesnik RP, Harryvan J, deVries JHM & Katan MB (1997): Fatty acids in serum cholesteryl esters as quantitative biomarkers of dietary intake in humans. Am. J. Epidemiol. 145, 1114–1122. | PubMed | ISI | ChemPort |