Primary Prevention of Breast Cancer

One might think that guidance regarding primary prevention would come principally from an understanding of the mechanism of breast cancer carcinogenesis. The answer to the question, what causes breast cancer, should lead to answers to the question, how can it be prevented. However it is not that simple! DNA damage and cell matrix damage are two postulated mechanisms for cancer initiation and proliferation. This involves spontaneous mutations and as well DNA damage from a number of sources, which cause breast cells to behave in abnormal ways, some of which lead to tumor formation, invasive behavior, and metastasis.

A search of the medical literature reveals numerous mechanistic studies probing the details of some aspect or other of breast cancer carcinogenesis, but there is a notable absence of any unifying theory, even in the area of hormonal involvement [2]. However, high blood levels of various estrogens have been clearly implicated as a risk factor for breast cancer. This conclusion is supported partly by a pooled analysis of nine prospective studies where blood was collected from healthy, cancer-free postmenopausal women who where then followed for an average of 5.4 years. The higher the levels of circulating estrogens, the higher the risk. While this does not prove a causal relationship, the authors suggest that other clinical and cell culture evidence makes the causal relationship quite likely [3]. The estrogen connection has recently been the subject of a comprehensive review [4].

Mainstream medicine for the most part views primary prevention of breast cancer as the use of pharmaceutical drugs, ovary removal or bilateral mastectomy, but restricts these interventions to those deemed high-risk, for example because of a genetic predisposition. Other preventive actions are outlined in the American Cancer Society (ACS) 2006 prevention guidelines as presented on their website are of interest: have several children and breast feed them for several months, don't drink alcohol, exercise regularly and stay slim and follow the ACS guideline for early detection of breast cancer. The guidelines mention the prescription drugs tamoxifen and raloxifene for high-risk women. So if a woman has already had her children and the last one is beyond breast feeding age, and this includes the vast majority of women concerned about breast cancer, then these recommendations boil down to exercise, keep weight down and avoid alcohol, period. Screening for early detection, for example by mammography, is not primary prevention. It has as its goal the early detection of cancer and is aimed at decreasing breast cancer-specific mortality and increasing overall survival. The success of mammography in accomplishing these goals is still being debated.

This review will examine the evidence associated with these and other potential preventive actions a woman can take for the purpose of primary risk reduction. The ACS guidelines in fact leave quite a lot unsaid.

NON-MODIFIABLE RISK (AND PROTECTIVE) FACTORS

Some would consider risk factors over which one has no control to be merely of academic interest. However, knowledge of these non-modifiable risk factors is in fact important, since their presence can encourage a woman to undertake aggressive investigation and implementation of preventive actions and interventions. The following are generally recognized as the principal non- modifiable risk factors for breast cancer [2,5,6]. While some of these factors are in theory modifiable, this would only apply to women of child bearing age who might elect to breastfeed their children for extended periods, have a large number of children, etc. For most women concerned with breast cancer prevention, these are not really options.

• Age
• Age at menarche (first menstrual period)
• Age at first childbirth
• Age at the onset of menopause
• Height
• Strong family risk of prostate cancer
• Number of pregnancies including zero (nulliparous)
• Breastfeeding and its duration
• Family history, genetic and polygenic factors
• Benign breast disease found on biopsy

Breast cancer risk is very low (< 10/100,000) before age 25 and increases up to 100-fold by age 45. After menopause the age dependence varies according to location, with women in the US and Sweden for example experiencing a slower but continual rise to age 75, whereas in Japan where the risk is low anyway, the cancer incidence after 45 exhibits a plateau followed by a slow decline. All countries studied show the sharp increase during the reproductive years, a pattern that suggests the involvement of the reproductive hormones [6].

A younger age at menarche and first childbirth and a later onset of menopause all increase the risk of breast cancer. The risk also decreases with the number of children. It is widely believed that the lifetime exposure to endogenous (internally generated) reproductive hormones is an important factor in the etiology of the disease, and it has also in fact been demonstrated that mammary epithelial cell proliferation can be correlated with serum ovarian hormonal levels (e.g. estrogens) [4,6]. Estrogens and progesterone are involved in the control of cellular proliferation in the breast, and the risk of breast cancer associated with this increased rate of cell division is hypothesized to relate to a greater opportunity for the accumulation of random genetic errors (mutations) some of which could impact tumorigenesis. Early menarche and late menopause extend this exposure. The age at first birth has a dramatic effect on the relative risk. When the comparison is with women who have had no children (nulliparous), an age at first birth of 15 years yields a risk reduction later in life of about 60%. As the age of first completed pregnancy increases, the benefit slowly decreases until it disappears (same risk as nulliparity) at an age of 30-35 and then becomes an increasing positive risk factor until menopause. However, the benefits of early pregnancy appear only about 10 years after birth and in fact, the risk is enhanced during the first 10 years with the enhancement increasing with age. The mechanism associated with the benefits of an early first pregnancy is unknown but animal studies suggest that as the breast develops, growth includes an increase in the number of stem cells and differentiation from stem cells to functioning ductal and secretor cells, and this growth and development involves the possibility of mutations and malignant transformation. The important point is that the final differentiation occurs during the first pregnancy, leading to a small increased risk of malignant transformation in the short term, but also yields fully differentiated cells that have a much lower risk of transformation thereafter [2].

The role of breastfeeding as a cancer risk factor has been the subject of numerous studies. In a recent review, MacMahon points out that the association between breastfeeding and breast cancer risk is manifest only in areas where accumulated lifetime lactation exceeds 5 years. Thus in developed countries where the mean lifetime duration of lactation is < 9 months, the association, while potentially present, would be difficult to detect because of the small magnitude of the effect. Also, MacMahon comments that the very long durations of lactation in underdeveloped countries may be only a marker for other factors which also influence risk, such as under-nutrition [2]. However, lactation results in a substantial delay in reestablishing ovulation following a completed pregnancy, and this would decrease the lifetime number of ovulatory menstrual cycles, which would have the same effect as late menarche or early menopause, both known to reduce risk [7].

The connection between breast cancer incidence and above average height is curious. The mechanism responsible for the relationship is unknown, but it has been suggested that nutritional status, which might interact with genetic factors, is related to the gain in height in childhood and as well to an earlier onset of puberty [8].

Benign breast disease is also termed a precancerous condition [9]. Benign breast disease generally develops in the breast ductal system. Extra cells develop on the inner wall of a duct - termed hyperplasia. Some of these cells can transform into odd-looking cells, called atypia. Women with severe atypical hyperplasia have a 4- to 5-fold increase in the risk of breast cancer, and a family history of breast cancer increases this risk up to 9-fold [2] If the hyperplasia intensifies and fills the duct with odd looking cells it is termed ductal carcinoma in situ (DCIS). When the abnormal cell growth expands through the ductal wall into the outside space and tissue, it is described as invasive cancer. Intraductal carcinoma requires treatment, generally surgical removal. Benign breast disease is generally identified on biopsy, but before it becomes invasive there is generally no palpable growth and evidence is generally from mammography. There are a number of other breast problems, most of which do not pose risk of progression to invasive breast cancer. These include swelling, pain, tenderness, lumpy breasts (not isolated breast lumps), nipple discharge, etc. that have historically been described as fibrocystic disease. A number of professionals in this field think that this term should not be used and the heterogeneous conditions it describes are not diseases at all [9]. However, the term is still used in both books for the lay audience and in the scientific literature. It is generally thought that this collection of disorders or physiological phenomena carry little or no risk of cancer. However, this may not be true for isolated lumps that a woman can generally differentiate from what is called "lumpiness" and the isolated lump demands professional attention and can indeed be cancerous. Some isolated lumps form as the result of ductal carcinoma in situ becoming invasive and developing into an extra-ductal cancerous lump. Lumps can also be cysts which, in general, are non-cancerous as in general are lumps attributable to fibroadenomas. However, professional advice should always be sought regarding lumps, if for no other reason, simply to alleviate anxiety. An informative discussion of this subject can be found in chapters 4 and 6 of Dr. Susan Love's Breast Book [9].

Mutations in the tumor suppressor genes designated as BRCA1 and BRCA2 are responsible for 80-90% of all hereditary breast cancers, but only between 5 and 10% of all breast cancers are attributable to inherited genetic mutations. Women who carry these mutations have an increase in lifetime risk of breast cancer of between 35 and 84% by age 70 [10], which is roughly 10 times that of the general population. The presence of this gene mutation also dramatically increases the risk of ovarian cancer (10-50% increase in risk by age 70). Ovarian cancer is difficult to diagnose at an early stage with present technology and has a high mortality rate. Removal of the ovaries almost but not totally eliminates the risk of this cancer. One reason the risk reduction is not 100% is that there may be some ovarian tissue left behind during the surgery.

This brings up the issue of genetic testing. Recent guidelines from the US Preventive Services Task Force (USPSTF) [10] provide guidance as to when the risk of having the BRCA 1/2 mutations is high enough to justify the recommendation of genetic counselling and testing. These guidelines depend on family history and the following definitions are important: relatives of the first-degree are parents, siblings and offspring of the individual in question; relatives of the second-degree are grandparents and grandchildren, uncles and aunts, nephews and nieces, and half-siblings. Given these definitions, what is termed increased-risk family history is determined as follows and depends on the presence of any one of the following conditions: (a) two first-degree relatives with breast cancer, one of whom received the diagnosis at age 50 or younger; (b) a combination of three or more first- or second-degree relatives with breast cancer; (c) a combination of both breast and ovarian cancer among first- and second-degree relatives at any age; (d) a first-degree relative with bilateral breast cancer; (e) a combination of two or more first- or second-degree relatives with ovarian cancer regardless of age of diagnosis; and finally (f) a history of breast cancer in a male relative. Ashkenazi Jewish heritage entails enhanced risk and an increased-risk family history for this group is defined as any first-degree relative or two second-degree relatives on the same side of the family with either breast or ovarian cancer. In general, women with one, two or three or more first degree relatives with breast cancer have, compared to women who have no relatives with breast cancer, relative risks of 1.8, 2.9 and 3.9 respectively for eventually developing breast cancer [11].

In an editorial accompanying the publication of these guidelines [12], Burk points out that potential candidates for BRCA testing need to be aware of the following limitations associated with genetic testing and interpreting the results: (a) current testing is estimated to miss 12-15% of BRCA mutations; (b) other genetic causes of family related breast cancer are likely; (c) finding a gene variant of unknown clinical significance is estimated to occur about 13% of the time. While it is important to be aware of these limitations, they do not appear to be a significant deterrent to proceeding with testing if an increased-risk family history is present.

Omitted from these guidelines is the risk factor associated with a family history of prostate cancer. A recent study done in Sweden found that women with a family history of hereditary prostate cancer had a 1.58 times greater risk of developing breast cancer than a woman lacking such history. Hereditary prostate cancer was in this study defined as at least three first-degree relatives with prostate cancer or two affected first-degree relatives of age < 55 years (early onset). The risk of breast cancer was increased to almost a factor of 4 in women before the age of 65 years from families with a history of early onset prostate cancer [13]. Similar results were found in a French study [14]. There have been reports that carriers of BRCA2 mutations are at enhanced risk of prostate cancer [13], and a study involving Ashkenazi Jewish men also found the BRCA2 mutation associated with an increased risk [15].

The USPSTF takes the position that women who have increased-risk family history would benefit from genetic counselling that allows an informed decision about testing and preventive measures. Also, some high-risk women may want to know their status because it impacts on the cancer risk for their existing and future children and grandchildren. Thus test results can influence what are called fertility decisions. There is in fact evidence showing that BRCA testing affects family planning decision-making [16]. Technology exists that permits pre-implantation genetic testing and thus embryo selection, but there are a number of associated issues, both financial and ethical.

MODIFIABLE RISK FACTORS

ESTROGEN METABOLITES IN THE ETIOLOGY OF BREAST CANCER
The estrogens estrone and estradiol are both metabolized to yield a number of intermediate and end products. Interest in these estrogen metabolites as possible factors in breast carcinogenesis can be traced back to at least the early 1970s. By the early 1980s it was found that women with breast cancer had elevated levels of one of the metabolites, 16a-hydroxyesterone. In a study published in 1997 which had a significant impact on research in this area, Kabat et al reported on a case control study [17] where the emphasis was on the ratio of two metabolites, 2- hydroxyestrone and 16a-hydroxyestrone (2-OHE and 16-OHE). There were only 39 cases and 58 controls, and for the total group, no significant association with breast cancer risk was found for the ratio of these two metabolites. However, for postmenopausal women (23 cases vs. 28 controls) a significant increase in incidence of breast cancer was found for low values of the ratio of 2-OHE/16-OHE, but the confidence limits, while not including the null result, were huge. There was no control for alcohol or coffee consumption. While this one statistically significant result based on a very small number of cases eventually became a key part of the folklore of estrogen metabolites, even the authors suggested that until large prospective studies were conducted, the connection between this ratio and the risk of breast cancer would remain uncertain. Also, a serious problem with this and some subsequent studies was that individuals with cancer were compared to controls, whereas if the interest is in primary prevention, then this ratio should be examined in prospective studies where the participants are cancer-free at baseline.

After Kabat et al there appear to be only four prospective studies that relate to this question [18- 21]. Two of these studies gave results that were statistically insignificant, one found that high, not low values of the ratio of 2-OHE/16-OHE were associated with increased risk, and one, which used mammographic breast density as a surrogate measure, also found high, not low values of the ratio were related to enhanced risk. Thus while cell culture evidence suggests that 16-OHE is pro-carcinogenic, has proliferative effects and down-regulates apoptosis (programmed cell death) and 2-OHE does not or only weakly influences these functions [22], epidemiologic prospective studies fail to confirm with consistent, statistically significant results the importance of the ratio of these metabolites in connection with the risk of cancer in initially healthy women. This view dismisses several suggestive case-control studies simply because they involved cases already diagnosed with breast cancer.

It is unfortunate that when one applies modern standards for evidence, the above-discussed studies by and large fail to achieve significance and also yield contradictory results. The reason this is unfortunate is that this metabolite ratio can be modified by dietary or supplemental intervention. The supplement in question is indole-3-carbinol, a compound found in such vegetables as broccoli. Supplementing with I3C or its first metabolite diindolylmethane (DIM) will generally shift the 2-OHE/16-OHE ratio to higher values, i.e. the production of 2-OHE is favored. This is an important aspect of Dr. E. J. Conley's protocol described in The Breast Cancer Prevention Plan for cancer prevention [23] and DIM is part of Dr. J. McWherter's protocol described in the recent book Avoiding Breast Cancer [24]. While health benefits may result from taking I3C or DIM or eating lots of broccoli, such actions do not appear justified for breast cancer prevention by prospective cohort studies even if there is a shift in estrogen metabolism in favor of 2-OHE.

TOTAL FAT
For several decades the conventional wisdom connected total fat intake with breast cancer risk. Walter Willett in his book Nutritional Epidemiology provides a review of the historical data and discusses possible sources of confounding [25]. An international study published in 1975 showed a strong linear relationship between national total dietary fat intake and the death rates attributable to breast cancer. However, as Willett points out, one of the potential confounders, adult height acting as a surrogate for energy balance during development, produces almost the same international correlation as fat intake. As more and more studies were reported, it became evident that there was in fact no connection between breast cancer incidence and total fat consumption. Prospective follow-up studies are generally regarded as the most reliable in ascertaining risk [25] since they avoid recall bias and bias in control selection inherent to the case-control approach. Hanf and Gonder from the University of Goettingen have recently (2005) reviewed the evidence [26]. Of 12 cohort studies which examined total fat intake, the only statistically significant result was that in one study increasing total fat decreased the risk. Nine prospective studies of meat (red and white) failed in a pooled analysis to reveal any risk. And animal fat was not found to be significantly associated with risk. However, as discussed below, one study found in premenopausal women that animal fat, mainly from red meat and high fat dairy food, was in fact a risk factor. Aside from this one result, the conclusion based on a large number of prospective cohort studies is that there is no significant evidence that total fat intake is connected with breast cancer risk [27-29].

It is also significant that no large interventional study has demonstrated that a low-fat diet reduces the risk of breast cancer [26]. Consistent with this, the recently reported results from the Woman's Health Initiative (WHI) Randomized Controlled Dietary Modification Trial of postmenopausal women found no connection between the low-fat dietary intervention and reduced risk of breast cancer [30]. However, in spite of the extensive media attention, this study can hardly be considered definitive since by the end of the sixth year of follow-up the intervention and control groups differed by only 8.1% in the energy intake from fat and perhaps more importantly, adherence to the dietary program in the intervention group declined to 31% at year 6 and 19% at year 9. While the overall measures of benefit from the low-fat intervention failed to reach significance, women with an initial very high dietary fat consumption as a percentage of total energy (> 36.8%) at enrollment had a significant reduction in breast cancer risk, a result that merits further study.

One of the problems with many of the studies of breast cancer and total fat intake is that they involved only a small representation of premenopausal women. In 2003 research from Harvard based on the Nurses' Health Study II was reported which addresses this issue [31]. Dietary fat intake and breast cancer risk were assessed among over 90,000 premenopausal women between the ages of 26 and 45 at enrollment (mean age 43), starting in 1991. Food frequency questionnaires were used at baseline and again in 1995. After 8 years follow-up, the relative risk of breast cancer based on total fat intake was slight and only marginally significant (RR = 1.25, 95% confidence limits 0.98 to 1.59). However, the incidence of breast cancer associated with the intake of animal fat exhibited, for increasing quintiles compared to those in the lowest quintile of consumption, a significant trend of increasing risk (RRs = 1.28, 1.37, 1.54, 1.33, PTREND = 0.002). An analysis of food groups revealed that animal fat, red meat and high-fat dairy foods were each associated with increased risk. The connection with animal fat may relate to cooking practices which may introduce carcinogens, and high-fat dairy foods contain fat-soluble hormones and/or growth factors, which may be related to breast cancer risk. Thus the connection with fat may be indirect. These interesting results apply only to premenopausal women, a factor that may explain the discrepancies between these results and all other studies, where the subject populations were mostly or entirely composed of postmenopausal women.

SPECIFIC FATS
Fat can be classified according to saturated, monounsaturated, polyunsaturated, and trans-fat types. Many studies lump all of these together and simply examine risks relative to total fat. Saturated fats have been traditionally one of the subtypes highlighted for disapproval in the context of breast cancer. However, as Hanf and Gonder [26] point out, nine of ten prospective cohort studies failed to find a connection and one found a reduced, not elevated risk. Monounsaturated fats such as found in olive oil have traditionally been regarded as protective, but 6 of 9 studies found no protective effect, two detected elevated risk, and one showed the expected lowering. Polyunsaturated fats (omega-3 and omega-6 fats) such as those found in plant oils and oily fish were examined in 8 studies with all results of no statistical significance.

Japan is a special case due to the low risk of breast cancer in the general population. In a recent large prospective cohort study [32] of women 40-79 years of age with a mean follow up of almost 8 years, a significant decrease in risk was found for the highest vs. the lowest quartile for fish and long-chain omega-3 fatty acid intake (RR = 0.56, 95% CI 0.33-0.94 and 0.50, CI 0.30-0.85 respectively). Among postmenopausal Japanese women at baseline, the highest quartile of vegetable fat intake was associated with a 2.08-fold increase in risk (95% CI 1.05-4.13). Consistent with this study, the Singapore Chinese Health Study reported significant reduced risk of breast cancer associated with dietary omega-3 fatty acids from fish and shellfish. Relative to the lowest quartile of intake, individuals in the highest three quartiles exhibited a 26% reduction in risk. While omega-6 fats were not in general implicated in risk, for subjects who consumed low levels of marine omega-3 fatty acids, a comparison between those in the lowest vs. highest quartiles of omega-6 fats had almost double the risk of breast cancer. Consistent with other prospective studies, no risk was associated with either total or saturated fat. A recently reported prospective study from Sweden also identified vegetable oil-based dietary fats, which are high in omega-6 fatty acids, as presenting a statistically significant risk for breast cancer (Odds Ratio = 1.74, 95% CI 1.12-2.72). Thus studies that lump omega-3 and omega-6 fats together and simply look at total polyunsaturated fats may be uninformative.

These results suggest the importance of the omega-3/omega-6 dietary ratio on breast cancer risk. Simopoulos [33] has reviewed the relevant literature concerning this ratio and cancer risk. While much more research is needed to clarify the role of the polyunsaturated fatty acids and in particular the omega-3 to omega-6 ratio in the context of breast cancer, the above results are very suggestive. It is interesting in this context that the typical North American and European diets result in a very low ratio of omega-3 to omega-6, both in the diet and in the cellular phospholipids (in blood cells) [34]. This is due to both a very high consumption of omega-6 fats and low consumption of omega-3 fats. Thus attempts to associate omega-3 or fish intake and breast cancer risk in countries in these areas may be doomed from the start because of the overwhelming influence of the omega-6 intake. This may be why conflicting results have been reported concerning risk and fish intake in North America and Europe [26]. It would be interesting to study the long-term intervention with eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) supplementation on breast cancer risk where the omega 3/omega 6 ratio in serum cellular phospholipids was monitored. This ratio (actually the EPA/arachidonic acid ratio or its inverse) can now be measured with a blood test available from commercial laboratories and there is significant evidence concerning the optimum range of this ratio in connection with a number of important health issues. The ratio is easily modified simply by consuming fish oil or the purified omega-3 fatty acids EPA and DHA [34]. The reader is referred to the International Health News newsletter (March-May 2005) for a research report that provides a detailed discussion of the role of polyunsaturated fatty acids in inflammation, immunity and a number of other health issues.

DIETARY PATTERNS
Humans obviously do not generally eat individual foods in isolation but rather consume complex mixtures of nutrients at most meals. Thus in recent years there has been growing interest in attempting to identify favorable and unfavorable so-called dietary patterns based on epidemiologic studies aimed at establishing a connection with the risk of various diseases and disorders. Analysis actually involves recovering patterns a priori from large databases of food consumption where information exists concerning the presence of various diseases in the cohort in question. Dietary patterns are particularly interesting because breast cancer shows strong variations with location of residence and changes in risk take place that are associated with migration from low- to high-risk areas. This suggests that a component of risk involves lifestyle issues including diet.

In this context there have been five recently reported prospective cohort studies that examined the connection between dietary patterns and the risk of breast cancer [35-39]. Large cohorts from North America, Italy and northern Europe were involved. The results seem quite disappointing, with only a salad and vegetable pattern yielding significant results in the Italian study and a so-called Southern pattern showing reduced risk in a large US study. In the salad - vegetable pattern, the prominent components were raw vegetables and olive oil. The foods most strongly implicated in the Southern pattern were cabbage, and legumes. The southern pattern was associated with a group having a lower level of education and more manual labor type work. The authors suggest that for this particular group, the southern pattern may also have been the diet experienced during the developmental years, and this may play a role in its protective effect. Diet patterns such as the Prudent Pattern (higher intake of fruits, vegetables, whole grains, low- fat dairy products, fish and poultry) or the Western Pattern (higher intakes of red and processed meats, refined grains, sweets and desserts and high-fat dairy products) were not found to have any significant association with breast cancer risk. In addition, in the Italian study, a different Western Pattern (potatoes, red meat, eggs and butter), the Canteen Pattern (pasta and tomato sauce) and a different Prudent Pattern (cooked vegetables, pulses and fish) all were found not to be associated with increased or decreased risk. As Mannisto et al comment [37], the results of dietary pattern analysis support "the suggestion derived from traditional epidemiology that relatively recent dietary habits may not have an important role in the etiology of breast cancer."

A recent study by Fung et al [40] relates to this failure of diet pattern analysis. Instead of patterns, they looked at the closely related "quality score" of several diets. While no risk or benefit was found in the context of breast cancer prevention for postmenopausal women in general, stratification by tumor type found on prospective follow-up of a Nurses' Health Study cohort that scoring high on the items in certain diets correlated with risk reduction for estrogen receptor negative tumors. Since this tumor type is in the minority, these results would explain why pattern analysis studies and diet studies which did not or were unable to stratify for tumor estrogen receptor type typically yielded null results. The diets found to be potentially protective were heavy in fruits and vegetables, fish, whole grains, had a high monosaturated to saturated fat ratio, a low percentage of energy from fat or saturated fat, and one diet approximated the traditional Mediterranean diet. The traditional Mediterranean diet has in a number of studies been associated with reduced risk of cancer [41].

FRUITS, VEGETABLES AND FIBER
Hanf and Gonder [26] have recently reviewed the evidence for a connection between fruit and vegetable consumption and the risks of breast cancer. Prospective cohort studies have uniformly failed to reveal a risk reduction. Riboli and Norat [42] in a review and meta-analysis published in 2003 identified 15 case-control studies and 10 prospective cohort studies. When all studies were analyzed together, no significant effect was seen for either fruits or vegetables. The same conclusion was reached by Smith-Werner et al in a pooled analysis of 8 prospective cohort studies - consumption of fruits and vegetables was not significantly associated with reduced breast cancer risk. Likewise, a recently reported prospective study involving over 285,000 women with a follow-up mean of 5.4 years found no association with either total or specific vegetable and fruit intake and breast cancer [43]. These null results may also reflect the estrogen receptor problem discussed above.

Hanf and Gonder [26] have also reviewed five prospective cohort studies regarding the association of dietary fiber and breast cancer. Four of five found no association and to this can be added a study which was not included and which also gave a null result.

DAIRY PRODUCTS
Moorman and Terry [44] have reviewed the literature regarding the consumption of dairy products and the risk of breast cancer. They found that the available epidemiologic evidence failed to support a strong association between either milk or other dairy products and the risk of this disease. A pooled analysis of cohort studies by Missmer et al [45] also found no associations with dairy products. The related issue of vitamin D and calcium contained in dairy products will be discussed later.

CARBOHYDRATES, DIETS WITH HIGH GLYCEMIC LOAD, AND INTAKE OF HIGH GLYCEMIC INDEX FOODS
The widespread reaction to the notion that dietary fat was bad and at the root of many ailments resulted in a significant change in dietary habits in North America and an increase in carbohydrate consumption to compensate for the loss of calories from fat. Much of the carbohydrate involved was of high glycemic index, a measure of the impact of the carbohydrate on blood sugar levels. As discussed above, evidence currently available suggests that the condemnation of fat was supported by improperly conducted studies, was simplistic and ignored the multiplicity of fat types. It may take some time before this is generally appreciated. Hyperinsulinemia (high circulating insulin levels), which can result from prolonged consumption of high glycemic index carbohydrates, has been shown to be associated with increased breast cancer risk. Thus the interest in a potential association with diets having a high glycemic load (glycemic index times the amount consumed) from large amounts of rapidly digested carbohydrates. While retrospective and case-control studies have suggested a positive association between high glycemic index or high glycemic load diets, seven prospective cohort studies reported over the last three years and involving a total of 368,272 women provide little support for this association [46-52]. In the face of massive null results, the few significant associations that have been reported [29,53,54] may possibly be merely due to chance in spite of statistics that suggest otherwise.

DIET AND NUTRITION IN EARLY LIFE
Studies looking back into childhood not only have the serious problem of distant-time recall, but also the nature of the foods involved has changed, making it difficult to apply the results from studies that look back into the past. For example, the trans-fat content of many processed and so-called fast foods is currently decreasing dramatically due to pressure from governments and the public and the unfavorable publicity being given to fast-food vendors who have not as yet addressed this problem. Also, drinking whole milk is now uncommon, and the "unnatural" constituents of milk (hormones, antibiotics, growth factors, various pesticides, etc.) have changed over the years. The contemporary practice of shortening the interval between pregnancies in cows has altered the endogenous bovine estrogen content of milk. Long-term recall is also highly problematic for both mothers who may be quite old when surveyed regarding the diets of their now adult female children during early childhood through adolescence, and women themselves may have paid little attention to what they ate during this period except to recall the foods they disliked. The sort of detailed quantitative recall necessary for a potentially informative study may in fact be impossible. The few studies that have been reported will not be reviewed. Any positive or negative judgments concerning the impact of current childhood diets on breast cancer risk 20- 50 years from now seems too theoretical.

CONCLUSIONS

Risk factors that were described as non-modifiable appear to offer little opportunity for risk reduction. It would seem highly unlikely that a woman would base the decision as to when to have a first pregnancy on its impact on the risk of breast cancer, and the nature of modern society in fact frequently encourages delaying this event until after the completion of education and the establishment of a career have taken place, and intentionally childless marriages are not uncommon even though this increases the risk of breast cancer. Nutrition and lifestyle changes that might delay menarche do not appear to have been seriously investigated. Women considering breastfeeding their children should consider the benefits of prolonged breastfeeding on the risk of breast cancer in later life, but again many aspects of life in modern societies work against the practicality of such action, although it is certainly not impossible. Benign breast disease and conditions classified as fibrocystic disease, which may evolve into cancer, may be reversible. This is discussed later.

While there have been a large number of studies concerning breast cancer and diet, it has become clear that a number of important issues remain unresolved. In particular, insufficient attention has been directed toward premenopausal women who appear to be more sensitive to dietary content than the postmenopausal women who make up the majority of subjects in studies. Thus premenopausal women should perhaps worry about high intakes of animal fat, red meat and high-fat dairy products. Also, fruits, vegetables, fish, whole grains and a diet with a high monosaturated (e.g. olive oil) to saturated fat ratio may be protective even if there is no evidence for this from studies that fail or are unable to stratify by tumor estrogen receptor status. The same can be said for the Mediterranean diet. The diets found by Fung et al to be advantageous for premenopausal women are generally regarded as healthy diets.

Attention should perhaps be paid to the balance between omega-3 and omega-6 fatty acids. One school of thought maintains that some if not many of the health problems associated with modern industrialized societies involve excessive consumption of omega-6 fatty acids from such sources as vegetable oils which overwhelm the beneficial effects associated with the modest consumption of omega-3 fatty acids from, for example fish. Achieving a better balance of these two classes of polyunsaturated fats may be beneficial in the context of this review, and readers are encouraged to investigate this potentially important subject by referring to the sources mentioned above.

Finally, studies on individual dietary components such as fat and dietary patterns are generally corrected for the influence of energy balance and weight gain. Thus there are dietary issues that involve both fats and carbohydrates, which concern just the impact on weight and body fat distribution. The relationship between breast cancer risk and body fat distribution, weight, and weight gain at various stages in life will be discussed in Part II, as will beverages, which some would include among dietary factors. Also, for postmenopausal women, the absence of clear dietary directions for breast cancer risk reduction should not discourage interest in a healthy diet aimed at reducing the risk of heart disease, other cancers, diabetes, and elevated weight or obesity. The reader is referred to Dr. Walter C. Willett's book, Eat, Drink and Be Healthy, The Harvard Medical School Guide to Healthy Eating for evidence-based information concerning healthy diets [55].

ALCOHOLIC BEVERAGES
In sharp contrast to the parade of null results associated with the link between diet and breast cancer, for over two decades there has been a constant flow of convincing evidence in favor of the hypothesis that alcoholic drinks, even in rather small amounts, carry an enhanced risk of breast cancer. This body of evidence includes a number of prospective cohort studies. Dumitrescu and Shields [1] have very recently reviewed this subject. Significant points drawn from this review and other sources cited are as follows:

• The proportion of breast cancer attributable to alcohol (ethanol - ethyl alcohol) consumption among women in the US is 2.1%, i.e. about 14,000 cases per year out of about 667,000 cases. In Italy it is estimated to be as high as 10%. The figure for the US population is small because of the modest association between alcohol and breast cancer coupled with the generally low to moderate average level of alcohol intake among US women. A campaign to reduce alcohol consumption would thus have a small impact on overall breast cancer incidence in the US. In fact, a campaign for abstinence might increase overall mortality because of the beneficial effects of moderate alcohol consumption on cardiovascular disease risk [2]. For that small minority who consume large amounts on a more or less daily basis it appears to definitely be a risk issue [3].
• A collaborative analysis of 53 studies revealed a 32% increased risk (95% confidence limits 19-45%) for women with an intake of 35-44 g/day of alcohol, and a 46% (33-61%) increase risk for more than 45 g/day. One drink typically contains between 10 and 15 g of alcohol, and the normal-sized bottle or can of beer contains about 14 grams. Another way of summarizing the data is that the relative risk of breast cancer was found to be increased by about 7% for each additional 10 g/day intake of alcohol, i.e. for each daily extra drink over and above a threshold of about 10-15 grams of alcohol. Finally, the data can be viewed as follows: the cumulative incidence of breast cancer by age 80 years is estimated to increase from 8.8 cases per 100 women who are abstainers to 10.1 per 100 for those who consume 2 drinks per day and to 11.6 per 100 women who consume 4 drinks per day. The cumulative incidence by age 50 over this range from abstinence to 4 drinks per day changes from about 1.5 cases per 100 to about 2 cases per 100 woman [4]. The relative risks given above tend to perhaps exaggerate the absolute risk. Also, the risk associated with one drink per day containing 10-15 g of alcohol does not appear to be statistically significant [5]. Thus drinking two or more cocktails to relax at the end of the day or two or more glasses of wine at dinner become a significant risk factor only if habitual to the point of being more or less a daily practice.
• Daily consumption of more than 15 g of alcohol throughout life is associated with a 33% increase in risk [6].
• Overall lifetime consumption rather than over any specific time of life may be more strongly associated with breast cancer risk although studies are in general inconsistent [6]. A study concerning whether or not frequent binge drinking among high school and college age women increases the risk in later life would be of considerable interest given the enhanced susceptibility to some risk factors during this age period.
• The consistency of study results is remarkable considering that they span diverse geographical populations, different age groups, varying levels of consumption and different study designs [6].
• Alcohol has been found to cause breast cancer in studies with mice and rats.
• While the mechanism of carcinogenesis due to alcohol is unknown, several plausible mechanisms exist including the perturbation of estrogen metabolism, mutagenesis by acetaldehyde, a metabolite of ethanol, oxidative damage and a potential influence on various metabolic pathways that involve folic acid.
• Alcohol consumption is associated with increased breast density as seen on mammography in both pre- and postmenopausal women. Increased mammographic breast density is a recognized risk factor for breast cancer with a fourfold to six fold increase in risk.

Observations in the late 90s that the risk of breast cancer due to alcohol consumption appeared to be modified by folate/folic acid intake resulted in considerable research into this subject. This is a very important aspect of breast cancer prevention since near or total abstinence for everyone is unrealistic, but taking one folic acid pill daily or being sure of adequate dietary intake has the potential of being both acceptable and effective. Also, for women of childbearing age, having an adequate folate status is recognized to be very important in connection with avoiding a folate- related birth defect, and taking folic acid after pregnancy has become evident is generally too late since the defect (neural tube) occurs very early in the fetal development. Thus it is of interest to examine the evidence. Note that folate and folic acid will be differentiated. As will be discussed below, the former is found in unfortified food, the latter is a synthetic chemical which is used in both supplements and food fortification.

If we restrict our attention to prospective cohort studies concerning the folate/folic acid - alcohol connection published between 2002 and 2006, there are six with a total of almost 270,000 women followed for an average of about 13 years [7-12]. Four of these studies provided quite strong and statistically significant results favoring the protective effect of dietary and supplemental folate/folic acid on the risk of breast cancer among women consuming more than one alcoholic drink per day. One, with a short follow-up, did not support the protective effect of folate/folic acid on breast cancer risk [7], and one found a favorable result, but the outcome was just shy of statistical significance [10]. The studies with favorable results did not stratify for menopausal status or were restricted to postmenopausal women.

If one attempts to average over total folate/folic acid intakes that were protective in these studies, the amount is at or above 350-400 micrograms/day. In a recent study involving 25,400 U.S. women [13], Stolzenberg-Solomon et al found folate from food for the first to fifth quintile ranged from 336 to 473 micrograms/day (mean quintile values, after mandatory fortification increased the dietary intake), and total intake from food and supplements ranged from 335 to 1210 micrograms/day. Only about 8% obtained amounts exceeding 412 micrograms/day from natural and fortified food, but 37% obtained between 307 and 412 micrograms/day from this source. Thus eating a diet heavy in folate/folic acid rich foods can provide an intake similar to that found to neutralize the effects of alcohol consumption, but in the cohort studied by Stolzenberg- Solomon et al, 39% would have needed supplements to bring the daily intake above 307 micrograms/day.

However, the folate/folic acid matter is somewhat more complicated than generally recognized. Folic acid, as found in supplements and also added to foods as part of government mandated fortification, is a synthetic chemical not found in nature. Its great merit is that it is both cheap and stable and solves what is probably an impossible problem of extracting commercially significant amounts of folate from natural sources. While folic acid turns out to have more bioavailability than folate from food, only a limited amount is actually metabolized, and some of the remainder appears as free synthetic folic acid in the circulation system. The long-term effects of circulating unmetabolized folic acid are unknown, but alarming reports have appeared in the medical literature. In fact, concerns about the possibility of high serum levels of unmetabolized folic acid were already being raised as early as 1997 in response to fortification and the use of high-dose folic acid supplements (> 1 mg/day) in order to decrease the risk of neural tube defects [14]. Confirmation that the concern may be justified came in part from a study published in 2004 by Charles et al [15]. A group of 2928 women taking either a placebo, or 200 or 5,000 micrograms/day of folic acid for neural tube defect prevention were followed from the 1960's to 2002. For women randomized to the high doses of supplemental folic acid the risk of death attributable to breast cancer was twice as great as those on a placebo or low dose. This was a small study and the result could well have occurred by chance, and this study must be taken merely an indication of a problem with high does of folic acid. Incidentally, 5000 micrograms/day (5 mg/day) is no longer used for this purpose. Nevertheless, 5000 micrograms/day would, on the evidence of earlier and later work [14,16], definitely elevate serum levels of unmetabolized folic acid.

In a study sure to add to concern associated with this matter, Troen et al [16] have just reported that unmetabolized folic acid in plasma is associated with a reduced natural killer cell cytotoxicity in postmenopausal women. Women who consumed a folate-rich diet and used folic acid supplement at a dose of > 400 micrograms/day had reduced natural killer cell cytotoxicity compared to those consuming a low-folate diet and no supplements, and unmetabolized folic acid in the plasma was a biomarker for excess folic acid consumption. This is of concern because such a decrease in cytotoxic activity compromises the immune function. Natural killer cells also play a role in tumor cell destruction and may be part of the first-line host defense against carcinogenesis. While the actual mechanism associated with the increased breast cancer risk that appears to be associated with moderately high to high intakes of the synthetic chemical folic acid is unknown, this provides a plausible biological explanation for the alarming experimental and epidemiologic results.

Related to this issue is the paper already cited above in which Stolzenberg-Solomon et al [13] reported on a study supported by the National Institutes of Health wherein it was found that high intakes of folate/folic acid actually increased the risk of breast cancer both in abstainers and those who consumed alcohol. To achieve high intakes, supplements were necessary. Neither folate/folic acid from foods alone nor natural folate from unfortified foods were significantly associated with increased breast cancer risk, perhaps because of the limitation on daily intake. But women consuming 400 micrograms/day or more of supplement derived folic acid had a 19% higher risk of postmenopausal breast cancer compared to those not taking supplements. Those in the highest fifth of the study population in terms of total folate/folic acid intake had a 32% greater risk as compared to those in the lowest fifth. These results which were statistically significant, were based on a large sample size and were corrected for confounding. The levels of folic acid intake found in this study to enhance breast cancer risk were, as discussed above, in the range known to increase the probability of unmetabolized folic acid in the circulation. However, Stolzenberg-Solomon et al do not mention unmetabolized folic acid as a possible factor in their discussion of possible mechanisms. Nevertheless, taken together with the results of Charles et al and Troen et al it would seem that warning bells are ringing.

It is evidentially quite easy to find individuals taking 400 micrograms/day or more of folic acid in a multivitamin or B-complex pill or even a separate folic acid supplement and as well eating enough folate-rich foods and fortified foods to push the total folate/folic acid up to 600-800 micrograms/day or even higher, an amount that would produce unmetabolized folic acid [14,16]. Troen et al [16] found unmetabolized folic acid in the 78% of plasma samples from the cohort in his study. Thus it follows that until more is known about the risks of synthetic folic acid intake, caution is indicated. The common "B-50" B-vitamin formulation contains 1 mg/tablet and Life Extension's "Two-Per-Day" tablets contain an 800-microgram dose. Supplements containing only 200 micrograms per minimum dose are readily available and may merit consideration. The typical amount in a multivitamin is 400 micrograms per pill, which is just at the threshold of potential danger according to Stolzenberg-Solomon et al for a woman consuming a diet low in folate with little by way of enriched food products. Someone who has a high consumption of fortified cereal, bread and products made from fortified flour may have cause for concern regarding the high intake of folic acid and would thus want to question the amount of additional folic acid also being consumed in supplements, including that from a simple multivitamin. In this case, 400 micrograms per day of a supplement may be excessive. A point to remember is that only folate from unfortified food appears safe at high intake, since all other sources are a synthetic chemical which is only metabolized to a limited extent and is responsible for the unmetabolized folic acid found in the blood of individuals ingesting large amounts of this chemical. The benefits of folic acid in the context of neutralizing the effects of alcohol consumption disappear, according to Stolzenberg-Solomon, at an intake exceeding about 400 micrograms per day and at higher intakes increased risk appears associated with an intervention intended to reduce risk. Having to deal with hidden sources of folic acid in fortified food only complicates this matter, especially when a single serving of breakfast cereal can provide 400 micrograms of folic acid in a form identical to that found in supplements.

There are other issues as well. In the last few years there has been increased interest in the relationship between risk factors, preventive actions and the tumor hormone receptor status, i.e. the histological estrogen receptor positive (ER+) or estrogen receptor negative (ER-) nature of the breast cancer that develops. How this estrogen receptor status influences the risk enhancement of alcohol and the risk reduction brought about by folate/folic acid is of interest because ER- tumors typically occur in only 20-35% of cancer cases. Thus there are two critical questions. First, is there an association between the adverse effect of alcohol and the tumor type that eventually occurs? Second, is the protective effect of folate/folic acid dependent on the histological type of tumor that occurs, i.e. does folate/folic acid protect against the occurrence of both histological types? As regards the first question, the literature is inconsistent with some studies finding no difference and others finding exclusively one or the other histological type influenced by alcohol consumption. One of the largest and most recent studies, which used data from the Nurses' Health Study and looked for risk factors for breast cancer according to estrogen receptor status, found no differential association with alcohol as a risk factor [17]. An earlier study found no differences for post menopausal and heterogeneous results for premenopausal women [18]. Other studies have been inconsistent [19,20].

In connection with the second question, a recently reported large prospective study [12], again based on data from the Nurses' Health Study, found that higher folate intake, when the alcohol intake was greater than 15g/day, reduced the risk of developing only ER- breast cancer. Stratification according to total folate for alcohol consumption greater than 15 g/day produced only one statistically significant result which was for folate = 534 micrograms/day. Another study also found protection only for ER- cancer, but for some obscure reason, the highest level of alcohol consumption was equal to or greater than 1/4 drink a day with no stratification for much higher consumption, if it occurred at all [8]. The result that the effectiveness of folate/folic acid is restricted to ER- tumor formation is puzzling since folate/folic acid intake has been observed in numerous studies to reduce the risk of breast cancer all the way to that seen in abstainers. If folate/folic acid is only beneficial for a small percentage of the cohort in question, i.e. those who go on to develop ER- tumors, one would think that folate/folic acid would not have been found to be that successful. This, incidentally, is why the question of whether or not alcohol promotes both ER+ and ER- cancer is important. If it is confirmed by other prospective cohort studies that the alcohol induces both ER+ and ER- tumors, and that folate/folic acid inhibits the alcohol effect only for those that go on to develop ER- tumors, there would be serious implications simply because folate/folic acid intake would then provide no protection from the alcohol promoted development of ER+ tumors, which represent the majority of those observed. Given that a woman has no a priori knowledge of the histological tumor type that might develop, this result would imply that the only preventive action in this context is limiting alcohol to no more than one drink a day on average. Using folate/folic acid for prevention would in fact produce a false sense of security when consuming larger amounts of alcohol.

Additional studies related to the folate - ER+/ER- question appear to be urgently needed, especially since the result suggesting that only ER- cancer is influenced by folate is based, essentially, on only one significant and relevant result bobbing is a sea of non-significant results. Women who drink on average more than 15 g of alcohol per day (more than about 1 drink) may want to give consideration to optimizing their foliate/folic acid just on general principles. Others may elect to play it safe and keep their consumption such that it averages out to less than one drink per day until this critical uncertainty is resolved.

GREEN TEA, GREEN TEA EXTRACT AND BLACK TEA
Tea intake is estimated to be second only to water in terms of worldwide consumption as a beverage [21]. Cell-culture and animal studies have shown that green tea and green tea polyphenols have anticarcinogenic properties against breast cancer [22,23]. However, there has been only a limited number of prospective cohort and case-control studies. The most recent meta-analysis of the cohort studies yielded an indication of a decrease in breast cancer risk for the highest vs. the lowest intake levels, but the result was not statistically significant. When a single case-control study was added to the analysis, the protective activity of green tea became statistically significant [21]. Similar less than definitive results were obtained in an earlier meta- analysis with a slightly different set of studies. Definite conclusions appear impossible considering the small number of studies, the lack of a consistent dose-response, and the absence of clinical intervention trial evidence. Typical intakes that were suggestive of benefit were in the range of 5-7 cups a day. Similar intakes can be achieved with green tea extracts, but for this option only cell culture and animal studies are available, although these generally provided evidence of potential benefit [22,23].

In a just published prospective cohort study from Japan, it was reported that green tea consumption was associated with reduced mortality due to all causes and due to cardiovascular disease, but not with reduced mortality due to cancer. Stratification of the cancer results did not include breast. The cardiovascular benefits were stronger for women than for men and appeared even at a level of consumption of 1-2 cups/day [24].

There is no statistically significant evidence concerning the merits of black tea in this context. Case-control and cohort studies are conflicting. In a recent meta-analysis, eight case-control studies yielded only a minor inverse (protective) relationship for black tea consumption and five cohort studies showed a modest increase in breast cancer risk [25]. However, for one specific type of breast cancer, i.e. lobular, black tea afforded a significant reduction in risk [25]. In general, studies of coffee and tea in the context of breast cancer risk have not examined the risk according to the estrogen receptor type of tumor that is involved or develops.

COFFEE
Epidemiologic studies of the association between coffee consumption and breast cancer risk have yielded inconsistent results. However, a recent case-control study that included large numbers of both pre and post menopausal women has just been reported [26]. A large and statistically significant beneficial effect of coffee consumption was seen with = 4 cups/day of regular coffee (Odds Ratio 0.62, 95% CI 0.39-0.98), but only for premenopausal women. In another study just reported [27], the effect of coffee consumption was studied in women who carry the BRCA mutations which put them at high risk. In this multi-center case-control study, a strong and statistically significant protective effect was found for BRCA 1 cancers. The protective effect was not seen until the daily consumption reached a level of 6 cups of caffeinated coffee, but at this level of consumption, the benefit becomes impressive and statistically significant (multivariate Odds Ratio = 0.31, 95% CI 0.13-0.73). Only caffeinated coffee reduced risk, and a significant risk reduction was not found for women diagnosed after the age of 50. The mechanism of this protective effect, if it indeed exists, is unknown, but the authors suggest phytoestrogens in coffee as a potential explanation. Also, Baptista et al [28] have suggested that gene-specific deactivation of various pathways involved in carcinogenesis could be promoted by substances in coffee. More research is clearly needed.

WEIGHT
Adult weight gain, both prior to and after menopause as well as postmenopausal obesity has in many studies been associated with the risk of postmenopausal breast cancer. There are also some studies that address the association of body fat and premenopausal breast cancer risk. In the past two years a number of studies have been reported which reinforce earlier studies and provide additional information. Among the issues involved are the use or non-use of hormone replacement therapy (HRT), the estrogen receptor status of the cancers found on follow-up, the presence or absence of hereditary predisposition (the BRCA mutation status), and when the weight gain occurred that ultimately influenced the risk. The highlights of these recent studies are as follows:

• Weight gain over the period from age 18 to menopause caries a significantly enhanced risk of postmenopausal breast cancer risk, with the risk increasing with the amount of weight gain.
• Weight loss or weight maintenance during these years reduces the risk [29-31].
• Among women who did not use postmenopausal hormone therapy, 24.2% of breast cancers could be attributed to a weight gain of 2 or more kg (4.4 lbs) since age 18 and 7.6% attributed to this weight gain since menopause [30].
• Hormone replacement therapy modifies the risk associated with adult weight gain. Women who experience adult weight gain and do not take hormone therapy have a greater weight- associated risk than those on hormone therapy [30,31].
• The association of breast cancer risk and adult weight gain appears to be restricted to those women who develop estrogen receptor positive tumors [31,32]. This is the most common tumor type.
• Postmenopausal obesity (BMI = 30, BMI - weight in kg divided by the square of height in meters) or being overweight (BMI 25-29.9) were associated with statistically significant risk of breast cancer in non-users of HRT [32]. Weight loss in early adult life (ages 18-30) protects against early onset BRCA-associated (hereditary) cancers [33] and weight control through dietary energy intake restriction is a justifiable action for risk reduction for these high risk individuals [34].
• Postmenopausal women with the highest energy intake, highest BMI and least physical activity had twice the risk of breast cancer when compared to those with the lowest energy intake, lowest BMI and greatest physical activity [35].
• For women who never used HRT, the sustained loss of 10 kg or more of weight since menopause was associated with substantial and significant reduction in the breast cancer risk (relative risk 0.43, 95% CI 0.21-0.86) compared to those who maintained weight after menopause [30].
• Compared to women with stable weight (2 kg - 4.4 lbs) during adulthood, a long-term weight gain (measured from age 20) of 15-20 kg (33-44 lbs) resulted in a 50% increase in breast cancer risk during the postmenopausal years, but only among those not currently using HRT. Long-term weight gain did not increase the risk of breast cancer occurring during the premenopausal period of life. [36].
• High birth weight, high stature at age 14, low BMI at age 14, and peak growth at an early age (rapid growth between 8 and 14) were found in a recent study to all be associated independently with increased risk of breast cancer [37].
• The risk of premenopausal breast cancer decreases with increasing BMI and greater body fatness during childhood and this association is independent of adult BMI [38].

Eliassen et al comment in their paper [30] that given the poor success of sustained weight loss after menopause, women should avoid weight gain throughout adult life rather than count on losing weight after menopause. One of the most noteworthy aspects of the above results is the interaction between risk, weight gain and postmenopausal hormone use. A simple explanation of this phenomenon is that while adipose tissue is a source of circulating estrogen, the addition of exogenous hormones (not made internally), i.e. HRT, obscures the influence of enhanced hormone levels from fat tissue due to weight gain or obesity. The use of hormone therapy to negate the adverse effect of postmenopausal weight is obviously not a solution, given the increased risk of breast cancer now recognized to be associated with this intervention.

Finally, in an editorial, Michels and Willett [39] from Harvard comment that currently available data suggest the following description of a lifetime body build with the lowest breast cancer risk: "one would want to be born light, grow slowly but steadily into a chubby, short child, and to maintain one's fat mass until one reached menopause, at which point one would want to shed the excess pounds immediately in order to keep the risk of breast cancer low." This quote highlights the complexity of the matter under discussion. Michels and Willett also point out that lower BMI during adolescence is related to a lower risk of cardiovascular disease and diabetes. Thus encouraging a relatively high BMI during adolescence may be undesirable from the standpoint of overall health.

SMOKING
In 2002 Terry and Rohan published a review of the literature concerning smoking and the risk of breast cancer [40]. At that time, the results were inconsistent although there was some indication of increased risk associated with long-term use of tobacco. In 2006 Cui, Miller and Rohan updated the subject with a further review of the literature and an update on their prospective cohort study [41]. Eight new studies had been reported. Three case-control studies and all the prospective cohort studies (three also) found increased risk for breast cancer associated with long-term cigarette smoking. In addition, two case-control and five cohort studies have examined the association between smoking before the first full-term pregnancy and breast cancer risk. All of the case-control studies and four of the five cohort studies found smoking before the first full- term pregnancy increased the risk of breast cancer. In the authors' prospective study, the highest breast cancer risk was found among women who smoked 40 years or more (RR 1.50, 95% CI 1.19-1,89). Positive associations were found with smoking duration, intensity, and cumulative exposure. They also observed that the lower the age at which smoking commenced the higher the breast cancer risk.

Thus there appears to be fairly strong evidence that breast cancer can be added the list of adverse outcomes associated with long-term heavy smoking, a list which includes lung cancer and cardiovascular disease. One would hope that such a dismal report card, including the added risk associated with smoking prior to first pregnancy, would discourage teenage girls from taking up the so-called "habit" but it is probably true that this group does not scare easily. Obviously any woman who smokes should consider the evidence presented above as providing a strong incentive for quitting.

EXERCISE
It is well known that exercise and physical activity have a beneficial effect on many aspects of health. In the context of breast cancer, most of the attention has been directed at the question of exercise and its intensity during adolescence and early adulthood and how this impacts the risk of in later life. Lagerros et al [42] have performed a meta-analysis of 23 studies (4 prospective cohort and 19 case-control) that provide considerable insight into this question. Overall, exercise during adolescence and young adulthood resulted in a significant 20% reduction in later breast cancer. This result was significant for women diagnosed after menopause and was almost significant for premenopausal cases. Both moderate and vigorous levels of physical activity were equivalent. It is also possible that adolescent physical activity carries over into adult life. There is also some recent evidence suggesting that increased physical activity even in the postmenopausal period can reduce the risk of breast cancer [43].

A population susceptible to this alteration in light exposure is the shift worker, and thus it was natural to look for enhanced disease risks in cohorts such as nurses and other health care workers, airline personnel, etc. In connection with breast cancer and shift work, a recent meta- analysis of 13 studies is of considerable interest. Studies included were of the following design: prospective cohort, nested case-control within a prospective study, retrospective case-control, or incidence studies where the referent group was the general population. Studies involved flight attendants, nurses and other individuals whose occupation involved night work. The aggregate estimate for all the studies was a statistically significant 48% increase in occupational risk of breast cancer. The results were similar when stratified for the type of occupation. Residual confounding did not appear to be significant. The fact that the risks for flight attendants and other night occupations were essentially the same was taken by the investigators to indicate that increased radiation exposure experienced by flight attendants is not a factor [44].

Two of the studies [44,45] were of prospective cohort design and were based on the Nurses' Health Study and the Nurses' Health study II, the former involving mostly postmenopausal and the latter premenopausal women. Increased risks of breast cancer were 39% in the former and 79% in the latter (reference 15 in [44]) both statistically significant, when shift workers were compared to those who just worked days. For the premenopausal cohort, the increased risk was apparent only after 20 years of rotating shift work whereas for the other study, it was after 30 years.

While the biological mechanism for the above observations is unknown, a popular theory involves the suppression of secretion of the hormone melatonin brought about by the presence of light during the night hours. Humans have numerous vital biological processes that vary reproducibly over each 24-hour period. The general term applied to this phenomenon is "circadian rhythm." Light controls the circadian related processes, i.e. the body uses light to determine where it is in the 24-hour period. Presumably this involves responding to both light and darkness. The principal actor is the pineal gland which is responsible for the secretion of the hormone melatonin, a process which is triggered by darkness during the hours when one is normally asleep. Light can acutely suppress melatonin secretion and disrupt the human circadian system [46]. In fact, it has been found that low melatonin levels, as measured by a biomarker in first morning urine, is associated with a significant increase in breast cancer risk [47]. The acute suppression of melatonin secretion by light during the night would thus be expected to increase risk and help explain the night-shift results.

It is also possible that disruption of the circadian rhythm by light exposure during childhood and adolescence may affect the lifetime risk of breast cancer since these developmental periods may be times when the female is particularly vulnerable [48]. For now, this is just a hypothesis, but having children become accustomed to sleeping in total darkness seems reasonable, given the evidence presented above concerning shift work.

Night-shift work has also been identified as a risk factor for colorectal cancer in the Nurses' Health Study. Nurses working night-shifts for 15 years or more had statistically significant multivariate (adjusted for confounding) relative risk of colorectal cancer of 1.35 (95% CI 1.03 to 1.77) [49]. The relationship of night-shift work to other cancers does not appear to have attracted much attention. However, melatonin is under investigation for use in cancer treatment, either alone or as an adjunct to conventional therapy [50]. A detailed review of melatonin and cancer can be downloaded free from the Life Extension website (www.lef.org, January 2004 issue of the magazine Life Extension).

Thus the question - what to do? For many on shift work, there is generally no choice. It is part of the chosen profession such as nursing or being a flight attendant. No studies have been conducted or even appear to be in progress that address potential solutions to this problem. However, there is a large literature on the subject of improving the performance of night-shift workers by attempting to adjust their circadian clocks. The most frequently seen suggestions involve (a) having a room at work that is very brightly illuminated (i.e. mimics outdoor light intensities) where breaks can be taken; (b) wearing dark glasses for the commute home in the morning; and (c) going to bed at once in a totally dark room. Whether such a protocol would have any impact on the risk of breast cancer remains to be seen, but if the increase in risk requires 20- 30 years to develop, it seems unlikely that a study could be conducted. Another potential solution would be to take melatonin, which is available over-the-counter in some countries including the U.S. and Canada. However, no one appears to have studied the long-term effects of taking this hormone, even though it is thought to have a number of interesting properties including acting as an antioxidant and in cell culture studies, being an anti-cancer agent [50].

Exposure to relatively weak electromagnetic fields (EMFs) such as found in the residential or occupational setting have repeatedly been suggested as a risk factor for breast cancer. A biological basis was provided by the observation that EMFs could inhibit the normal nocturnal rise of melatonin levels. Feychting and Forssen[51] have recently reviewed the evidence (up to and including 2005) associated with both workplace and residential exposure to low-frequency (50-60 Hz) EMFs. Included were studies on electric blanket use, a commonly quoted example of residential exposure. From the sum total of evidence available the authors conclude that no significant increased risk is associated with low-frequency EMF exposure. However, Davis et al recently reported that residential nocturnal exposure to EMFs decreased a urinary marker for melatonin [52]. In addition Schernhammer and Hankinson [47] found a decreased risk of breast cancer associated with an increase in this marker in a case-control study nested in the Nurses' Health Study II. Unfortunately, these two studies used a different urine collection protocol which makes it difficult if not impossible to ascertain if the marker level changes observed by Davis et al are significant in this context since they were very small compared to the range of marker concentrations found by Schernhammer and Hankinson and in this latter study, significant risk reduction was seen only at levels of the marker indicating high nocturnal melatonin levels.

MULTIVITAMINS
If one searches the medical literature database PubMed with the linked keywords breast cancer and multivitamin in either title or abstract, one finds essentially nothing of interest in the past 10 years. This reflects the practice of focusing on individual micronutrients. This is unfortunate since there is considerable evidence of the benefits derived from taking a multivitamin in the context of prevention of many health problems. Micronutrient deficiencies are related to cancer risk and multiple deficiencies are common. The evidence includes the observation that micronutrient deficiencies can mimic radiation or chemical damage to DNA causing both single and double-strand breaks and oxidative damage or both [53,54]. The double- strand chromosomal aberration is a strong predictive factor for human cancer (for a detailed discussion of this subject and the evidence supporting taking a multivitamin/mineral supplement daily, see the research review in the International Health News newsletter of September- October 2004 titled A Metabolic Tune-up?? What Is This All About? which was inspired by the extensive research of Bruce Ames and coworkers at the University of California, Berkeley). This section will be devoted to individual micronutrients.

VITAMIN D [55]
Among the micronutrients, vitamin D has received the most attention. This interest partly derives from the results of a number of studies indicating that solar ultraviolet radiation (UV) exposure is associated with a reduced risk of breast cancer as well as other cancers [56]. Individuals with a high level of exposure will have the highest levels of vitamin D and its metabolite 25-hydroxy vitamin D (see the IHN Research Report on Vitamin D in the May and June 2004 issues for a detailed discussion of vitamin D and health). In a recent study by Grant [56], 12% of breast cancer deaths among white American women and 16.5% among black American women were attributed to inadequate exposure to solar UV radiation. Supporting evidence comes from studies of serum levels of 25-hydroxy vitamin D. In a recently reported meta-analysis of two studies [57,58], Garland et al found that women who consume 1000 IU of vitamin D in addition to the normal background amount consumed or generated per day had a 10% lower risk of breast cancer, and in addition, an intake of 2700 IU was estimated to yield a risk reduction of 50% for an individual weighing 70 kg (154 lbs [59]. These intakes vastly exceed those found in prospective or case-control studies of the relationship between vitamin D intake and breasts cancer, which is probably why such studies have been inconclusive (the top quartile or quintile of intake rarely exceeds >500 or >700 IU per day. An daily intake of 1000 IU is being commonly recommended by scientists working is this field [60] (see also the above cited IHN Research Report). While some might be concerned that 2700 IU per day would be toxic, a recent study by Heaney et al [61] found that even 10,000 IU per day had no adverse effects. It is well known that if one sunbathes until the skin just shows a slight pink, the estimated generation of vitamin D is equivalent to the oral intake of between 10,000 and 20,000 IU. It is also important to realize that in northern latitudes (>35 degrees-40 degrees N) the amount of UV that is active in producing vitamin D in the skin is low to negligible in the winter months and vitamin D deficiency is common among those living in these higher latitudes or the equivalent in the southern hemisphere. Sun exposure or even prolonged sunbathing in the winter in Boston or Edmonton does not generate significant vitamin D. In addition, individuals with dark or black skin need twice the exposure to achieve the same Vitamin D production. Also, the current recommendation to avoid all sun exposure unless a sunscreen is used has had an impact of deficiency levels, since sun screens decreases the natural generation of vitamin D. Thus lifestyle, attitude toward sunscreens, and where one lives all have a significant influence on vitamin D status, and, according to the study discussed above [56], the incidence of breast cancer. The National Academy of Sciences'sInstitute of Medicine has set 2000 IU per day as the safe upper limit for vitamin D intake (see [60] for a current discussion of toxicity and recommended intake).

Low vitamin D status is also implicated as a risk factor in colon cancer, colonic adenomas, prostate cancer and ovarian cancer. The evidenced in general comes from dietary, serum and geographical studies [60]. The impact of vitamin D deficiency in connection with bone health, hypertension, diabetes, and rheumatoid arthritis is discussed in the above-cited IHN Research Report along with the necessary information to interpret the standard blood test for vitamin D status (25-hydroxy vitamin D). In the study of Garland et al [59] the 50% risk reduction for breast cancer was achieved with blood levels of 25-hydroxy vitamin D > about 50 ng/mL (125 nmol/L).

The hypothesis has also been recently advanced by a group including eminent vitamin D researchers and epidemiologists that the remarkable and recurrent seasonal aspect of influenza is due to low vitamin D levels in the winter [62]. The authors of this study suggest that even 2000 IU/day in the winter may be insufficient to fully protect against influenza, especially for the elderly.

Women living in northern latitudes and those who avoid sun exposure clearly need to be concerned about their vitamin D status. In the context of intakes of for example 1000 IU per day, food sources are not significant, and at least during the winter months, supplementation with the D3 form of vitamin D (not the D2 form) appears to be necessary. Whether supplementation is necessary in the summer is obviously a strong function of the total unprotected sun exposure.

VITAMINS A, C AND E
Aside from cell-culture studies, there appears to be no evidence of significance regarding the role of these three vitamins in breast cancer prevention and there is some evidence that intakes of vitamin C in excess of 300 mg/day was associated with an increase in risk [63-65]. However, Zhang has reviewed a large number of studies concerning vitamin C and breast cancer risk and reports no association [66]. If vitamin C increased the risk, presumable it would have been seen in at least some of these studies.

CALCIUM AND MICRONUTRIENTS IN DAIRY PRODUCTS
Two reviews published in 2005 both found in general no association between the consumption of dairy products and breast cancer [67,68]. However, a large cohort study restricted to postmenopausal women published in 2005 found that dietary calcium and/or some other component of dairy products may modestly reduce the risk of breast cancer in postmenopausal women, an effect that was stronger with estrogen receptor positive cancers [69]. Also, another prospective cohort study found that for premenopausal women, high intakes of low-fat dairy products were linked to lower breast cancer risks, but it was impossible to separate the effects of vitamin D and calcium [70]. This last mentioned study is consistent with the result that vitamin D and calcium from food and supplements was associated with lower levels of mammographic breast density which is a surrogate for lower breast cancer risk [71]. However, in this study as in others, the intake of vitamin D and calcium were tightly correlated, making it difficult to identify independent associations.

Thus the independent role of calcium is unclear in this context, and if dairy products are protective, the association is weak and the role of both estrogen receptor and menopausal status uncertain.

IODINE AS A PREVENTIVE AGENT IN BREAST CANCER
The breast and thyroid are the principal accumulators of iodine. Venturi [1] and Cann et al [2] have hypothesized that dietary iodine deficiency is associated with breast pathology and cancer, and others have also discussed this hypothesis [3,4]. The evidence is as follows (see [1] and [2] for references):

• Clinical studies indicate that treatment with iodine reduces or eliminates the symptoms of some forms of benign breast disease. This is significant because some types of benign breast disease carry enhanced risk of developing breast cancer.
• The progression from intraductal hyperplasia to intraductal hyperplasia with atypia and then to ductal carcinoma in situ (see Part I) appears to be favored by low iodine status and reversed by iodine supplementation. Hyperplasia with atypia and intraductal carcinoma in situ carry an enhanced risk of developing invasive breast cancer [5].
• Traditional Asian medicine has long used iodine-rich seaweed to treat various types of benign breast disease, e.g. to soften breast tissue and reduce breast nodulation.
• Enhanced iodine accumulation in the breast occurs during pregnancy and lactation. It is common to find decreased breast tissue density and nodulation following pregnancy and lactation.
• Iodine reacts with fat to form iodolipids which are thought to be involved in the regulation of proliferation of breast tissue. Deficiency would lead to enhanced proliferation which could contribute to both benign breast disease and cancer.
• Thyroid dysfunction related to iodine deficiency is seen in some breast cancer patients.
• Iodine absorption occurs in the same ductal epithelium where the majority of breast cancer originates.
• Iodine is considered a prerequisite for normal breast tissue development in both animals and humans and in animal studies it has been shown that a deficiency results in benign abnormal tissue growth, malignant tissue growth and an increased sensitivity to carcinogens.
• Japanese women have the lowest breast cancer incidence in the world and have an average intake of about 12 mg/day of iodine from iodine-rich foods such as seaweed, products derived from seaweed, and fish. Some sources quote the Japanese intake as high as 45 or more mg/day. In countries where breast cancer is high, the iodine consumption is in the range of 0.1 to 0.2 mg/day, and may even be lower among women who avoid salt, the principal dietary source after fortification was mandated to combat an epidemic of goiter. It is important to note that
• Japanese women who live near the sea and eat lots of fish and seaweed products have an enhanced incidence of gastric cancer, and it is thought that this is due to very high levels of iodine intake.
• In rodents, iodine inhibits or delays induced carcinogenesis and reverses the pathological changes produced by iodine deficiency (cystic changes, periductal fibrosis and lobular hyperplasia).

There is clearly considerable circumstantial or anecdotal evidence in the above observations and a noteworthy lack of human studies and clinical trials, especially ones aimed at testing the hypotheses that iodine deficiency is a breast cancer risk factor and that supplementation with iodine will decrease the risk of breast cancer in North American or European women. However, two clinical trials discussed below address the question of treating some aspects of what some call fibrocystic disease. As discussed in Part I, fibrocystic disease is not generally benign breast disease of the type that carries a risk of developing into breast cancer, and there is considerable support for not using the term since it encompasses a variety of conditions, most of which are not really considered diseases. Nevertheless, the term continues to be used. The extent to which the problems covered under the umbrella term fibrocystic disease increase the risk of developing true benign breast disease or breast cancer over the long-term appears unknown.

With regard to the two trials, one used as an endpoint the elimination of breast pain and/or the reduction of micronodular growths, tenderness, fibrous tissue plaques and macrocysts [6]. The use of iodine therapy produced significant positive benefits for the majority of participants in the trial, which was multinational and placebo controlled. In the second study the endpoint was the elimination or reduction of pain from what is called cyclic mastalgia, which is breast pain that correlates with the phase of the monthly period [7]. Again, significant improvement was found with iodine therapy. The observed benefits of iodine therapy found in these two trials may or may not extrapolate to reduce cancer risk. However, in his book Avoiding Breast Cancer [8] McWherter describes a Canadian study where 3000 women living in Ontario were given an iodine preparation and followed for 10,000 woman-years. The incidence of breast cancer was half the rate of women in the same age bracket who did not take the iodine supplement. It does not appear that this study was published. McWherter uses iodine supplementation in his clinical practice at the FEM center in Texas where he specializes in breast care and the treatment of benign and malignant breast disease.

McWherter [8] also describes a small in-house study of iodine status where a single "loading dose" of iodine/iodide was given to nine consecutive breast cancer patients and urine samples collected. All showed deficiency as judged by the level of urinary iodine/iodide excretion which was significantly below what was considered optimum.

The formulation of the iodine supplement may be important. There is evidence that molecular iodine (I2) is appropriate for breast tissue whereas the iodide ion as in potassium iodide is active in the thyroid. Supplementation increases the loading of both breast and thyroid tissues, and large doses must be given to increase the load in both. Typical therapeutic doses are 6 mg/day. One convenient source is Lugol's Solution, which is a mixture of molecular iodine and potassium iodide and provides approximately this dose in one drop. SSKI, a saturated solution of potassium iodide contains 19-50 mg/drop. The proprietary formulation called Iodoral contains 12.5 mg of iodine/iodide per tablet in the same proportion as Lugol's solution. McWherter mentions in his book that he uses Lugol's solution in treating benign breast disease. It takes a number of months to see benefit [8]. Lugol's solution is available over the Internet and at some compounding pharmacies.

It is probably safe to say that nobody knows what the optimum daily intake of iodine is. In the US, the average daily intake is estimated at about 0.25 mg/day. Table salt is fortified with potassium iodide (KI) or potassium iodate (KIO3). In the US, this has raised the iodine status in low-iodine regions (the so-called goiter belt) to the point where goiter is uncommon (but not unknown). It is also generally agreed that iodine deficiency can cause hypothyroidism, mental retardation, and cretinism (severe mental retardation accompanied by physical deformities). Fortification of table salt has also resulted in decreased incidence of cretinism. However, many households do not use iodized salt, and salt avoidance is common because of concerns over hypertension. This leads one to suspect that the average consumption is just enough to prevent the deficiency diseases mentioned above from becoming noticeable, but the daily intake is very low compared to, for example, the intake by Japanese women. Those who have been active in research in this area regard a daily intake of a few hundred micrograms (few tenths of a mg) to be way too low for optimum health.

The circumstantial and anecdotal evidence presented above should be sufficient to prompt human studies designed to examine the iodine-breast cancer hypothesis. The potential benefits could be huge, but the safe limit of iodine intake would be an aspect of such studies, given the possible connection between very high intakes and gastric cancer, although the intakes responsible for this risk enhancement would appear to vastly exceed therapeutic doses used in the context of breast disease and would, no doubt, never be used in studies. Endocrinologists who consider one mg/day excessive and dangerous will raise concerns about supplemental iodine causing either hypothyroidism or hyperthyroidism, although there does not appear to be any significant evidence that a dose of 6 mg/day will cause either disorder except under unusual circumstances. The study on the relief of cyclic mastalgia used a proprietary preparation called IoGen at doses up to 6 mg/day for 6 months without evidence of any adverse effects [7]. The formulation is made by Symbollon Pharmaceuticals, a company which is now enrolling women for a one-year, FDA-approved randomized trial of 6 mg/day of iodine derived from taking IoGen for periodic breast pain associated with symptomatic fibrocystic breast disease. It is significant and instructive that those who believe doses of one mg per day or more are dangerous were unable to block this US government approved study. For more information regarding this trial and a list of participating sites in the US, consult http://clinicaltrials.gov/show/NCT00237523. Individuals wishing to try iodine supplementation should consider doing so under the supervision of a physician.

Space does not permit a more detailed discussion of the potential role of iodine in breast cancer prevention. The reader is referred to an article available free online by Dr. Donald W. Miller, MD, professor of cardiovascular surgery at the University of Washington (http://www.lewrockwell.com/miller/miller20.html).

SOY AND SOY ISOFLAVONES
Interest in the hypothesis that soy foods and the phytoestrogens they contain reduce the risk of breast cancer extends over several decades [9]. Soy is a rich source of isoflavones which have chemical structures similar to estrogen and under certain experimental conditions bind to estrogen receptors and exert estrogen-like activity.

Three recent reviews of the literature are available [9-11]. The most ambitious which is by Trock et al concludes that while soy intake may be associated with a small reduction in breast cancer risk, the results must be interpreted with caution due to the potential of exposure misclassification, confounding, and lack of dose response, and in addition there are some experimental results suggesting adverse effects [11]. Gikas and Mokbel simply conclude that "there is no clear evidence that phytoestrogen intake influences the risk of developing breast cancer" [10] and Messina et al report the conclusion of a recent workshop -- more and better research is needed [9]. Even the American Soybean Association recently withdrew their petition to the FDA that, if successful, would have allowed the claim that there was an inverse association between soy protein intake and breast cancer risk [12].

ORAL CONTRACEPTIVES
The use of reproductive hormones as contraceptives began in 1960 and since then an estimated 200 million women have used them. Since most types of hormonal contraceptive contain an estrogen as well as a progesterone component, it is not surprising that questions would be raised regarding the possible impact on breast cancer risk. In a 1996 review (frequently referred to as the Collaborative Group Study or the Oxford Study), 54 studies were available for analysis which involved over 53,000 women with breast cancer and over 100,000 who were free of breast cancer [13]. These studies included data for oral contraceptive (OC) use by all age groups and a wide range of duration-of-use. This review concluded the increased risk ranged from 24% in current users to 16% 1-4 years after stopping, and 7% 5-9 years after stopping. No evidence of risk was found for those having cancer diagnosed 10 or more years after discontinuing OC use or diagnosed in women over 45 years of age. In this study there was no duration-of-use effect, no pill-type effect and no effect of age at first use. The results were statistically significant.

In contrast to the 1996 Collaborative Group Study, Marchbanks et al [14] found in a large case- control study reported in 2002 that for women diagnosed between 35 and 64 years of age, current or former OC use was not associated significantly with increase breast cancer risk, nor were OCs implicated in risk associated with a family history. An important difference between this study and the Collaborative Group Study was that the latter included women who were younger. Consistent with the Marchbanks et al study, a recently reported large French follow-up study found no increased breast cancer risk associated with prior OC exposure once a woman reached menopause [15].

Because young women appear to be more susceptible to breast carcinogenesis than older women, there has been considerable interest in the question of OC use in the teen years, the period leading up to the first pregnancy and the period prior to the onset of menopause. The following recent studies are of interest:

• A Swedish case-control study published in 2005 and limited to women diagnosed prior to age 41 found that OC use before age 20 was associated with a 110% increase in risk, whereas OC use before having the first child carried an enhanced risk of 63%. For women diagnosed prior to age 36, there was a 53% increased risk per year of OC use prior to age 20. While each year of OC use prior to age 20 resulted in a significant increase in risk of early-onset breast cancer, there was no risk associated with use after age 20 [16].
• In a US study involving women 20 to 54 years of age, all premenopausal, for those developing cancer before age 35 recent use of OCs increased the risk by 126% [17].
• In another US case-control study the subjects were 20-44 years of age at diagnosis. For women < 35 years of age, the results when stratified by the dose of the estrogen component gave a risk of 262% for high vs. 91% for low dose [18].
• A large prospective study [19] with more than a 7-year follow-up conducted in Norway and Sweden enrolled women aged 30-49. Current or recent use of OCs at enrollment was associated with a 60% increase in breast cancer risk. In the 30-39 age group, the significant enhanced risk was 50%, 70% and 50% forever use, current use, or former use respectively, as determined at enrollment. Lower but significant risk was found for the 40-49 age group. This study found enhanced risk associated with use before first full-term pregnancy and also that long-term users were at higher risk.
• A meta-analysis (an analysis of studies) just published addresses the risk of OC use both during the premenopausal period (age < 50) and as well during the period prior to the first full- term pregnancy [20]. All of the cases in this analysis of case-control studies developed cancer prior to age 50. Compared to non-users, ever-users of OCs had a 19% increased risk of breast cancer. For women who had had one or more child, the increased risk was 29% in general and 44% if OCs were used prior to vs. 15% for use after the first pregnancy. Use of OCs for 4 years or more prior to the first pregnancy carried an enhanced risk of 52%. All of these results were found to be statistically significant.

Therefore, while one recent study cited above found no risk for OC use after age 20, the other studies did not confirm this conclusion and enhanced risk for use at younger ages was commonly found. Furthermore, these results lend credence to the suggestion that the disagreement between the 1996 meta-analysis and the study of Marchbanks may be due to the quite large risk factors associated with OC use by young women, an age group mostly absent from the latter study. These results with younger women are consistent with the enhanced susceptibility of carcinogenesis among this age group and especially among those who have yet to bear a child, and in addition, are consistent with the association between breast cancer risk and the magnitude and duration of estrogen exposure. Nevertheless, there is the widespread belief that OC use is benign in this context, a belief that may well be correct if a woman has not developed breast cancer for 10 years after cessation of use or has entered menopause. Unfortunately, most of the studies available involved older and frequently stronger formulations of the pill than are currently prescribed, and these modern formulations may carry reduced risk, as suggested by one study cited above. Whether these low-dose formulation will carry a small but significant risk will take a number of years to determine, given the apparent lag time between exposure and clinical presentation of this disease, the variety of formulations and modes of administration and the relatively recent introduction and use of these low-dose formulations. The one study cited above that relates to more modern low-dose formulations still found significant risk [18].

Thus there is a risk-benefit problem which may not even be fully appreciated by most women, and this problem goes well beyond the question of contraception. As Burkman points out in an interesting review published in 2001, oral contraceptives provide protection against ectopic pregnancy, reduce the risks of ovarian and endometrial cancer, protect against pelvic inflammatory disease, reduce the incidence of benign breast disease, provide relief from menstrual disorders, reduce the risk of colorectal cancer, improve bone mineral density and finally reduce the risk of rheumatoid arthritis [21]. On the other side of the ledger, aside from the potential increase in breast cancer risk, there is an increase in risk of venous thrombolism and of stroke due to blood clots. Thus this is far from a simple matter. For example, any intervention that reduces the risk of ovarian cancer is important given that this type of cancer is difficult to diagnose before it is too late to treat with more than palliatives measures. Also, avoiding unwanted pregnancy can be very important for some individuals and carry great weight in the decision making process if other methods of contraception are rejected for one reason or another. The decision is particularly challenging for someone who has yet to have her first full-term pregnancy or is a teenager, since as discussed above, the evidence of risk is particularly strong and consistent for this situation. Thus there is a considerable challenge for obtaining guidance, especially if one is associated with a medical culture where the 10-minute office visit is the norm!

HORMONE REPLACEMENT THERAPY AND RELATED ISSUES
A recent review [22] presents a question which puts the hormone replacement therapy (HRT) matter in clear perspective: "Why, for decades, since the mid 1960s, were millions of women prescribed powerful pharmacological agents already demonstrated, three decades earlier, to be carcinogenic?" Yager and Davidson in their review of estrogen carcinogenesis in breast cancer [23] briefly examine the early evidence. In a meta-analysis of 51 studies involving over 160,000 women published in 1997, it was found that use of HRT or ERT (estrogen replacement therapy) for more than 5 years was associated with a statistically significant 35% increase in risk of breast cancer. The recent results of the Women's Health Initiative study (WHI) which also found significant risk in HRT (but not ERT) should have come as no surprise. This study also reported increased risk of venous thrombosis, cardiovascular disease and stroke [24]. The WHI study finally had an impact, was featured big-time in the media and reduced the prescription rate for HRT dramatically, leaving women with few adequately tested options to deal with severe menopause-related symptoms. The surprising fact is that up until then there was evidently insufficient concern to frighten women or change prescribing practices, in spite of at least 51 studies already in the peer review literature that collectively waved a red flag. This is in fact such an interesting phenomenon that in June of 2004 a group of historians, epidemiologists, biologists, clinicians and woman's health advocates gathered to examine the causes and implications associated with the question posed at the start of this section [22].

It is important to realize that HRT in North America has in recent decades almost always involved estrogens derived from horse urine and progesterone-like synthetic chemicals, generally denoted by the term progestins, but frequently confused with and erroneously equated to natural progesterone produced endogenously. The great appeal to the "industry" of synthetic progestins is that they can be mass-produced, patented, blessed with regulatory approval, and aggressively marketed. These progestins are now under scrutiny as principal actors in increasing the risk of breast cancer [25]. The estrogens produced from horse urine, also termed conjugated and these chemically modified conjugated estrogens actually contain a mixture of nine different estrogens.

In sharp contrast to the HRT generally used in North America, in Europe natural estrogen (estradiol) and other sources of progesterone have been popular for a number of years, primarily due to tradition [26]. In France a widely used HRT involves transdermal or injected estradiol and micronized progesterone or progesterone derivatives not used in North America. In a French prospective cohort study with a follow-up of almost 6 years, it was found that HRT that used estrogen and micronized progesterone gave no increased risk of breast cancer, whereas combined therapy that used synthetic progestins gave results that confirmed the above discussed increase in breast cancer [27]. Micronized progesterone is made from yams or soy, has a molecular structure identical to human progesterone, and micronization enables steady, even absorption. It is available in North America. In another French study reported in 2002, similar results were obtained, i.e. there was no increase in breast cancer in a cohort where 83% used transdermal estradiol gel and a source of progesterone other than the synthetic progestin used in North America [28]. In both of these studies, the results were statistically significant. Two conclusions are evident. There appear to be HRT protocols using bio-identical estrogen and progesterone that do not increase the risk of breast cancer, protocols that have been used in Europe and in particular in France for some time. Also, these results should focus even more attention on synthetic progestins such as medroxyprogesterone acetate as potentially part of the problem with enhanced breast cancer risk.

The term hormone balancing is encountered in discussions of HRT as well as in the more general context of female health and ageing issues. This term generally implies the use of natural so- called bio-identical estrogens and progesterone (i.e. identical to those produced by humans) for HRT. This is a complex subject and physicians who specialize in this area attempt to optimize the use of hormones for each patient, taking into account menopausal status, presence or absence of ovaries and/or uterus, age, and plasma hormone levels. Proper and successful orchestration involves monitoring and adjusting dose levels, ratios of estrogens, etc. and monitoring plasma levels. The goal, of course, is to minimize the risks and maximize the benefits. This approach also permits taking advantage of the potential beneficial effects thought to be associated with natural progesterone [8]. It appears debatable whether or not sufficient information is available to permit ascertaining the breast cancer risk associated with various hormone balancing protocols commonly suggested since more hormones than just estrogen and progesterone are frequently involved. Nevertheless, the French studies provide some evidence of safety for natural based HRT using estradiol and micronized progesterone. Readers interested the subject of hormone balancing may wish to consult Avoiding Breast Cancer by J. McWherter, MD [8] and Hormone Balance by Carolyn Dean, MD [29]. These two books will provide a starting point and an introduction to this rather complex subject. Incidentally, some experts (probably most!) in this area discourage self-testing with saliva test strips followed by self- medication with over-the-counter hormone preparations, and recommend that any use of hormones should be done under the supervision of a physician with experience in this area. But there is a potential problem. Some physicians who include hormone balancing in their programs aimed at prevention and treatment of breast and menopausal problems report considerable trouble with poor accuracy and precision of commercial laboratory hormone assays and incorrect formulations of estrogen and progesterone products by compounding pharmacies. This may be an area worth exploring during consultation.

Hormone balancing in general can go beyond manipulating doses of bio-identical estrogen and progesterone. Two other commonly used substances are testosterone and dehydroepiandrosterone, better know simply as DHEA. The latter is available over-the-counter in the US. Testosterone and DHEA are termed androgens. Testosterone is secreted by both the ovaries and adrenal gland whereas DHEA and its sulfate are adrenal androgens which can be converted into testosterone. Testosterone can be further modified by aromatase-mediated chemistry to give estrogen. This is not the case for methyl testosterone, a form frequently used in combination with conjugated estrogen for the treatment of low libido or other problems associated with oophorectomy or menopause. Methyl testosterone is termed non-aromatizable. It can cause adverse changes in lipid profiles and some physicians do not recommend its use [30].

There have been a number of studies regarding the potential risk of testosterone, both endogenous and exogenous. A recent large, multicenter case-control study nested in a prospective study (serum levels were measured before breast cancer developed) found that the highest serum testosterone levels gave about double the risk when compared to the lowest levels. Subjects were postmenopausal and did not use HRT [31]. This result is consistent with and similar to that found by Key et al [32] in a reanalysis of 9 prospective studies which also used serum levels where it was found that when the highest vs. the lowest serum testosterone quintiles were compared, there was a 122% increase in breast cancer risk. For all women taken together, the risk increase was 45%, and there was little dependence on the time from measurement to diagnosis. All these results achieved statistical significance. Similar results were reported in 2005 based on data from the Nurses' Health Study [33]. In a recent paper, Lillie et al examined the question of bias in these studies and found that the association between increased testosterone levels and increased breast cancer risk is unlikely to be due to bias or the lack of adjustment confounding [34].

In a recent study from Harvard of combined estrogen (conjugated) and testosterone hormone replacement in postmenopausal women, a significant increase in breast cancer risk was also found [35]. Most of the participants had been given methyl testosterone. However, the addition of testosterone to conventional HRT (conjugated horse estrogens and synthetic progestins) also increases the risk of breast cancer, although the relative risks were somewhat lower than those found by Key et al for non-HRT users [36]. The studies regarding breast cancer risk based on serum levels that found enhanced risk were mostly with postmenopausal women. For premenopausal women, studies are limited. However, two recent studies found a positive association between blood levels of testosterone and breast cancer [37,38]. Noteworthy by their absence are studies designed to access the breast cancer risk associated for women undergoing testosterone only replacement intended to bring levels up to the normal range [30] and the study cited above where risk was found with testosterone replacement therapy, subjects used mostly conjugated estrogen and methyl testosterone and thus the investigation does not directly address the question, although it does point to danger associated with one protocol. The use of methyl testosterone also confuses the issue. Thus the significance of the observed breasts cancer risk associated with testosterone replacement therapy at physiologic levels remains unclear.

The enhanced risk associated with DHEA therapy for postmenopausal women is smaller than that found for testosterone (19-69%), but the data are more limited [31,32]. Page et al found no association with either DHEA or DHEA sulfate in premenopausal women [39]. Kaaks et al [31] caution that while adequate levels of DHEA (or its sulfate DHEAs) may contribute to better bone mineral density, well-being and libido without significant effects on endometrial tissue, the association with increased risk of breast cancer "strongly caution against the use of DHEAs for postmenopausal hormone replacement." This warning was based on serum level data. How this relates to DHEA therapy intended to treat low levels is unclear.

These results for testosterone and DHEA are fairly recent, and some physicians using hormone balancing which includes these two hormones may be unaware of this literature. The two above cited books ignore or downplay the possibility that the use of testosterone or DHEA might enhance the risk of breast cancer as does the 2005 position statement from the North American Menopause Society, which, while citing the study of Key et al discussed above, fails to tell the reader the results [40]. However, in the latest edition of her book Dr. Susan Love's Breast Book the author cautions against both the use of testosterone and DHEA because of the possible connection with increased breast cancer risk[41]. Obviously more research is needed.

Women with high endogenous androgen levels (so-called hyperandrogenic) probably are totally unaware that this situation exists. But if for some reason these levels were measured and found high, is there anything that can be done to reduce them? Otherwise this subject is rather academic. In fact, there has been considerable interest in this question in Italy. Two studies have reported decreases in serum levels of testosterone brought about by a dietary intervention which included reductions in total fat and refined carbohydrates, an increase in the ratio of omega-3 to omega-6 polyunsaturated fatty acids, and increased intakes of foods rich in dietary fiber and phytoestrogens [42,43]. Other studies are ongoing [44]. The same group involved in these studies has shown that elevated testosterone levels in treated breast cancer patients are strongly associated with an increased risk of recurrence, and that a dietary intervention similar to the one described above reduced the recurrence rate [45]. This result is consistent with a just reported study that found the metabolic syndrome was a prognostic factor for breast cancer recurrences [46].

Finally, an important area of concern and interest is the use of unopposed estrogen therapy (no synthetic progesterone) and the risk of breast cancer. A study from Harvard just recently published addresses this question with an analysis of data from the Nurses' Health Study. In this prospective study, 11,508 postmenopausal women who had had a hysterectomy and reported estrogen use were included in the follow-up. The cohort was later expanded to a total of 28,835 participants. Users of unopposed estrogen (conjugated) were found to be at increased risk of breast cancer, but only with very long term use (> 20 years) [47]. This result is consistent with that found in the Women's Health Initiative Study discussed above which involved a much shorter follow-up and found no enhanced risk. These results, although based on a limited number of studies, should provide some reassurance for women using this approach to deal with estrogen deficiency associated with a hysterectomy or menopause. They should however be concerned about continuing the therapy past 20 years.

SPECIAL CASE OF HIGH RISK IN GENERAL AND GENETIC RISK (BRCA 1/2) IN PARTICULAR
As discussed in Part I, high risk can refer to having the BRCA1 or BRCA 2 mutation, but is can also refer to the presence of genetic risk inferred from family history that may not be related to the BRCA genes. This high risk level may prompt the recommendation of pharmaceutical or surgical intervention. The pharmaceutical intervention generally involves tamoxifen, an estrogen receptor modulator used mainly in treating breast cancer rather than for primary prevention. Randomized, controlled clinical studies testing the efficacy of tamoxifen for primary prevention utilize cohorts judged to be at high risk but in fact at enrollment there was for most studies no a priori information available regarding BRCA status. Pooled analysis [48] of several studies found an overall reduction in the incidence of breast cancer in the treated vs. placebo groups of 38%. Risk reduction was seen only when the cancers found were estrogen receptor positive. In a preliminary report just published [49] which involves a trial at the Royal Marsden Hospital in the UK, 2500 high-risk women were randomized to either tamoxifen or a placebo. At the time of the report, 70 cases of breast cancer had been observed and they were equally divided among the placebo and tamoxifen groups, i.e. no benefit from treatment. It is not clear why. The Marsden cohort had a much higher percentage of individuals with high risk based on family history rather than reproductive history, the presence of benign breast disease, atypical hyperplasia, etc.

Because of the importance of the question, does tamoxifen reduce the risk of breast cancer in individuals carrying the BRCA mutations, for two of these studies the number of BRCA carriers was determined later. Unfortunately, in both studies, the number of carriers was too low to provide any meaningful results. One study involved 13,388 treated participants and yielded only 19 BRCA carriers in the 288 cancer cases that developed. Only cases were tested [50]. In the other, which involved the above mentioned Royal Marsden Hospital study, out of 70 cases only 4 were carriers [49]. Again, the numbers are so small that meaningful analysis is impossible. Thus, if a woman is found on genetic testing to be a BRCA carrier, there appears to be no statistically significant evidence, based on randomized, controlled clinical trials, that tamoxifen will reduce the risk of developing breast cancer, provided this is the only factor contributing to a high-risk classification.

The problem of side effects associated with tamoxifen chemoprevention is nicely summarized by Bergh [48]. Based on pooled studies, treatment of 14,192 (not a typo) women for five years prevented (or deferred!!) 132 estrogen receptor positive cancers. The "expense" in side effects when treated vs. untreated subjects were compared was that 53 vs. 22 developed endometrial carcinoma, 118 vs. 62 had a thromboembolic event (e.g. deep vein thrombosis or pulmonary embolism), and 59 vs. 39 had a cerebrovascular accident or stroke. Incidentally, tamoxifen also appears to increase the risk of cataracts. Not a comforting score card nor the profile of a benign therapy. Thus physicians who must advise BRCA carriers are in a difficult position with regard to risk vs. benefit. Also, as Powles points out [51], the clinical trials of tamoxifen for primary prevention in cohorts deemed high risk have treated very large numbers of women but only a few actually developed breast cancer in either the treatment or placebo arms. The big percentage differences in breast cancer cases between the treated and placebo arms may be impressive (e.g. 50%) but the actual number of cases in each arm is negligible compared to the total number of participants treated and thus exposed to the risk of serious side effects. In other words, the absolute number of breast cancer cases prevented was very small. Another problem associated with seeking guidance from clinical trials is that, compared to the long times associated with risk of developing breast cancer, clinical studies of chemoprevention that run only 5 years may not be very informative, but very long trials may not be operationally or financially feasible and also involve long-term exposure to the risk of side effects. These are issues that the BRCA carrier may wish to explore with her physician before agreeing to undergo prophylactic therapy with tamoxifen or related drugs.

The BRCA carrier does have more effective primary preventive options that are evidence based - bilateral mastectomy or the removal of both ovaries. The former is profoundly disfiguring with potential psychological problems, although breast reconstruction has progressed significantly. Bilateral mastectomy yields approximately a 90% reduction in risk. Removal of the ovaries (oophorectomy) yields a 56% reduction of risk of breast cancer for BRCA 1 carriers and perhaps a 46% for BRCA 2 carriers [52]. An important consideration is that the risk of ovarian cancer, which is considerably enhanced in BRCA carriers, is also reduced by about 90% by oophorectomy. Even better results were reported recently at the annual meeting of the American Society of Clinical Oncology. It was reported that removal of both ovaries and fallopian tubes reduced the risk of breast cancer by 70% [53]. But oophorectomy has its own set of side effects since it amounts to surgically induced menopause which, if severe symptoms develop, leads to the need for treatment, and childbearing now becomes vastly more complex. However, when confronted with an estimated lifetime probability of 60-80% for having breast cancer, some women do indeed elect one or both of these radical preventive measures.

Serious questions can also be raised in the case of young women as to the long-term effect of oophorectomy as it relates to the resultant estrogen deprivation. A study from the Mayo Clinic just published in the Lancet (Oncology) addresses this issue [54]. Researchers investigated the impact of prophylactic bilateral oophorectomy on mortality. It was found that women having this surgical intervention before the age of 45 years had a 67% increase in overall mortality compared to controls. This increase in mortality was seen mainly in women who had not received post-surgical estrogen replacement up to age 45. There was increased risk of non- cancer related mortality unless estrogen replacement occurred. Premature estrogen deficiency that was not compensated throughout the post-surgical period up to age 45 increased the risk of cardiovascular disease, osteoporosis, bone fractures and neurological diseases. There was also an increased risk of estrogen-related cancers despite oophorectomy, but this was attributed partly to preexisting conditions. The authors point out that current practice involves prescribing estrogen after women undergo prophylactic bilateral oophorectomy before menopause, but whether they receive treatment up to age 50 is unclear, and the results of the Women's Health Initiative trial have prompted a striking reduction in the use of estrogen alone as well as estrogen plus progestins for all ages. They suggest that the results of the Women's Health Initiative study might not apply to women with natural or surgical menopause before the age of 50 years, and that this practice should be reconsidered.

Finally, there is growing interest in more intensive screening as a partial solution to the high risk associated with BRCA carriers or those with a high-risk profile from family history and other factors. Intensive in this context generally means adding additional imaging such as MRI to mammography. To quote Dr. A.S. Whitemore of Stanford University School of Medicine [53], " Putting myself psychologically in the shoes of a young women with a BRCA 1 mutation, I certainly would get MRI screening of the breasts. I would be religious about it. I would not have a mastectomy; I would just get very good screening." An aspect of this philosophy is that one avoids the heavy costs in side effects associated with surgical or pharmacological risk reduction measures and simply takes the risk of cancer developing with the hope of early enough detection to permit successful treatment. A woman electing this approach should consider researching the extent of local expertise with this relatively new application of MRI and perhaps seek out a center specializing in the monitoring of BRCA carriers and others at very high risk, even if some travel is involved. This approach also allows one to postpone irreversible therapy while waiting for better non-surgical preventive measures to be developed, an option that may be very attractive to younger BRCA carriers.

RISK ASSOCIATED WITH RADIATION EXPOSURE
It is generally acknowledged that radiation exposure is associated with increased breast cancer risk and is cumulative, but at low doses the risk is usually boarding on negligible [55]. For example, enough is known about the dose vs. risk to estimate that even a number of ordinary chest x-rays would only yield a relative risk of about 1.02, i.e. a 2% increase in risk, which is close to insignificant [56]. However, the picture changes a bit when age is factored in. For example, the extensive use of x-rays to monitor children with scoliosis significantly increased the risk of breast cancer in later life [57]. The number of x-rays taken ranged from a few to over 70. Exposure started at an early age, mostly before 14. It is well known that the sensitivity to radiation in this context is high in this age group [55]. Increased breast cancer risk is also seen in individuals, frequently young, who are given high dose radiation for Hodgkin's lymphoma and as well as in others who receive therapy that exposes the breasts to radiation. The benefits presumably outweigh the risks from therapeutic radiation in such situations.

However, an interesting situation is presented by BRCA carriers. In a just reported multi-national study [56], the risk of chest x-rays was examined in a cohort study of 1,610 women who had the BRCA 1/2 mutation. Compared to the above-mentioned groups where cumulative radiation exposure was high, this cohort had exposure only from routine chest x-rays, which according to the author's estimate, put them at least a factor of 10 lower in exposure as compared to groups where enhanced risk from x-rays was found. In this cohort of BRCA carriers, any reported exposure to chest x-rays was associated with an increase in breast cancer, the risk was increased in women aged 40 or younger and in women born after 1949, and the risk was particularly high for those exposed only before the age of 20. These results were described as clinically significant, given the already high risk among BRCA carriers and the potential for a two- to three-fold increase in risk associated with chest x-ray exposure. These results also raise the issue of potential risks associated with mammographic screening which is often recommended on an annual basis for BRCA carriers starting at age 30.

ENVIRONMENTAL ESTROGENS AND COSMETICS
The role of estrogen in the etiology of breast cancer is well established and this raises questions concerning the potential contribution from cosmetics and many chemicals in the environment that can enter the human breast and may have estrogenic activity. For example, parabens (esters of p-hydroxybenzoic acid) widely used as preservatives in underarm cosmetics (deodorants) have been implicated. These chemicals have inherent estrogenic and other hormone related activity and absorption might explain the clinical observation showing a disproportionately high incidence of breast cancer in the upper outer quadrant of the breast, just the local area of frequent and long-term application. Parabens have also been found in human breast tumors [58,59]. The fact that many chemicals thought to carry potential risk tend to accumulate slowly over the years is a worrisome aspect as is the possibility that while individual levels of a given chemical may not be dangerous, a number can act together to produce significant risk. The list of chemical candidates is long. The bottom line appears to be that no one knows the nature or seriousness of the risk and that more research is needed. However, such research is very difficult and in most situations, also presents ethical issues which prevent direct human testing. A discussion of individual chemicals and the associated evidence of risk are beyond the scope of this review. The reader is referred to a very recent and comprehensive review [60].

ASPIRIN AND NON-STEROIDAL ANTI-INFLAMMATORY DRUGS IN BREAST CANCER PREVENTION
A large number of studies have appeared concerning the potential of aspirin and other non- specific non-steroidal anti-inflammatory drugs (NSAIDs, e.g. ibuprofen) and specific COX-2 inhibitors such as Celebrex for the primary prevention of breast cancer. Epidemiologic studies have given conflicting results. Four large prospective cohort studies published between 1996 and 2005 [61-64] with long-term follow-up involving over 390,000 women and over 9000 observed breast cancer cases found mostly no significant evidence of benefit from the use of aspirin, ibuprofen or other NSAIDs and one study found a modest increase in risk associated with long term use [63]. Only about half of case-control studies yielded evidence of significant benefit [65]. When very large cohort studies with long follow-up yield evidence of no benefit and case-control studies are inconsistent, it is hard to make a case for the benefits of the intervention in question. As regards the specific COX-2 inhibitors, a recent case-control study found significant benefit associated with Celebrex and Vioxx but only 10 cases were compared to controls in the analysis [66]. Given the potentially serious gastrointestinal toxicity associated with long-term use of NSAIDs and the well publicized cardiovascular risks associated with the use of COX-2 inhibitors and perhaps even some non-specific NSAIDs taken at high dose [67,68], it is not surprising that there appears to be a reluctance among experts to recommend the use of any member of this class of drug for long-term primary prevention of breast cancer [69]. It should be mentioned, however, that the evidence that NSAIDs appear to play an important preventive role in colorectal cancer seems generally acknowledged as is the role of inflammation as a factor in the initiation and progression of a number of cancers [70].

The target of both specific (COX-2) and non-specific NSAIDs are the cyclooxygenase enzymes (COX-1 and COX-2) which, among other actions, convert the omega-6 fatty acid arachidonic acid (AA) to prostaglandins. The COX- 2 derived prostaglandin PGE2 (a so-called eicosanoid) is inflammatory and related to cancer through activating cell proliferation and migration, angiogenesis (development of tumor blood supply) and inhibiting apoptosis (normal programmed cell death) [70]. Both specific and non-specific NSAIDs inhibit the production of this prostaglandin through enzyme inhibition. Another way to suppress AA derived eicosanoids such as PGE2 is through a low intake of its precursor linolenic acid, the dominant omega-6 fatty acid in food, and a high intake of omega-3 fatty acids, especially the long-chain acids EPA and DHA discussed above. Experimental data suggest that a ratio of dietary omega-3 to omega-6 fatty acids needs to be 1:1 or 1:2 in order to provide protection against the development of cancer through the manipulation of the AA tissue concentration [71]. In most Western countries the dietary omega-3 to omega-6 ratio ranges from 1:10 to 1:50. It has been suggested that failure to obtain consistent results in studies that examined the effect of high dietary intakes of omega-3 fatty acids (e.g. from fatty fish consumption) is due to the overwhelming effect of high levels of AA and thus the impossibility of reducing the omega-3 to omega-6 ratio down to say 1:2 [72]. The point is that some of the same end results produced by NSAIDs may be achievable, apparently without side effects, by adjusting the omega-3 to omega-6 balance, generally by increasing the consumption of fish and/or supplementation with eicosapentaenoic acid (EPA), one of the omega-3 essential fatty acids found in fatty fish and fish oil [73] and reducing the consumption of the linoleic acid (found in vegetable oils like corn, soybean, sunflower and safflower oil, in meat and in many prepared food products). Individuals who consume large quantities of linoleic acid, the omega-6 precursor of AA, may be unable to achieve a beneficial omega-3 to omega-6 balance simply by eating fish once or twice a week or, in many cases, every day. The measurement of the cellular AA/EPA ratio in blood cell phospholipids, a test mentioned above that is now available at some clinical and diagnostic laboratories, can provide guidance as to current status and progress in suppressing tissue AA and its conversion to pro-inflammatory prostaglandins [73]. Someday there may be studies that examine the relationship between cancer risk, the omega-3 and omega-6 dietary intake and the resulting cellular ratio of AA to EPA. This would eliminate a serious shortcoming in past studies mentioned above that examined the relationship between diet, inflammation and cancer. Barry Sears argues on the basis of extensive data that bringing the AA/EPA ratio down to 2 to 4 should have significant health benefits, especially with regard to inflammation-mediated illnesses, and the list of such disorders is long [73].

CONCLUSIONS

Fairly strong evidence has been presented concerning risk reduction associated with limiting alcohol consumption or taking folic acid supplements to counteract the adverse effects of drinking. Premenopausal women should perhaps be concerned about high intakes of animal fat, red meat, high-fat dairy products and all women should consider limiting foods rich in omega-6 fatty acids and eating more foods high in omega-3 fatty acids or even supplementing with EPA and DHA. Avoiding oral contraceptives during the teenage and early adult years also appears justified from the breast cancer risk standpoint, but this is clearly a complex issue. Vitamin D is perhaps the most important micronutrient in the context of breast cancer prevention as well as other health issues, and both prudent sun exposure and supplements appear worth considering. Iodine intake and status appear to be much more important than generally recognized, and iodine supplementation is part of the breast cancer prevention program some physicians use. Avoiding HRT and limiting ERT to a few years if symptoms associated with menopause are intolerable appear indicated, but so-called hormone balancing may turn out to be an attractive option. Other actions that are supported by evidence and can be described as potentially effective involve not smoking, making exercise a life-long habit, and taking very seriously the matter of weight loss after menopause as well as obesity in general. Evidence has been presented concerning the merits of sleeping in total darkness.

Finally, there is the special case of high-risk individuals, identified either because of strong family history indications or because genetic testing has revealed a BRCA mutation. Women in this position should consider seeking advice and perhaps consider undergoing surveillance from a physician or a clinic specializing in this area. The side effects of medical prevention, be it through drugs or surgery, are of such a magnitude that these interventions must be carefully weighed for risk vs. benefit. A second opinion seems highly desirable. The most conservative approach seems to be very active surveillance with multiple imaging techniques done at a center specializing in high-risk individuals, and in addition, frequent examinations. There is evidence suggesting that BRCA carriers should minimize x-ray exposure but this is incompatible with frequent mammograms. Life is rarely simple when dealing with the risk of cancer or cancer itself. For those with high risk of breast cancer, the risk of ovarian cancer is also a very serious issue which women in this category should insist be addressed with the best possible surveillance protocols available.

This review has concentrated on preventive aspects that are based on clinical or epidemiologic studies. Results that failed to achieve statistical significance by the usual criteria based on confidence intervals (limits) have by and large been ignored. It turns out that there are a number of actions a woman can take that are rather solidly evidence based. Much fascinating research involving cell culture studies has been omitted even though it gives a glimpse of potential future preventive interventions. Finally, screening by mammography and its relation to primary prevention has been omitted intentionally from this review. It is a complex and highly controversial topic which may be the subject of a future review. For now, the interested reader is referred to the comprehensive and up-to-date review by Ralph Moss, Ph.D. which can be purchased for a nominal amount at http://www.cancerdecisions.com