Types of dietary fat and the incidence of cancer at five sites

Types of dietary fat and the incidence of cancer at five sites

PREVENTIVE MEDICINE 19, 242-253 (1990) Types of Dietary Fat and the Incidence of Cancer at Five Sites STEPHEN D. HURSTING, M.P.H.,’ MARKTHORNQIJ...

885KB Sizes 0 Downloads 1 Views



19, 242-253 (1990)

Types of Dietary Fat and the Incidence

of Cancer

at Five Sites



Research Program, Fred Hutchinson Cancer Research Center, 1124 Columbia St., Seattle, Washington 98104

The specificity of a statistical association increases the Likelihood that it represents a causal relationship. In exploring the relationship between dietary fat and cancer, specificity applies both to cancer sites (outcome) and to component fats (exposure). In this study, Armstrong-Doll criteria were used to select female cancer incidence data for breast, cervix, lung, and colon, and male incidence data for lung, colon, and prostate for 1973-1977from 20 countries with reliable registry data. Truncated age-standardized rates were correlated with estimates of per capita disappearance of total fat and of saturated, monounsaturated, and polyunsaturated (total, fish o-3, o-6) fats in 19751977. Multiple regression analyses were standardized for estimated total calorie intakes and used to assess the association between each fat and incidence at each cancer site. Estimates of per capita dietary and crude fiber intakes were also included in the analysis. Total calorie intake was not associated with cancer at any site when controlled for total fat intake, whereas total fat intake was strongly associated with cancers of the breast, colon, and prostate even after adjustment for total calorie intake. Cancers of the lung and cervix were not correlated with dietary fat intake. Monounsaturated fat had no positive association with cancer at any site. Saturated fat was positively associated with incidence of cancers of the breast, colon, and prostate and polyunsaturated fat was associated with incidence of breast and prostate cancers but not colon cancer. Fiber intake, when included in the analysis, affected the magnitude of the fat-cancer correlations, particularly between total fat and colon cancer. Fish omega-3 polyunsaturated fat had a nonsignificant negative association with the cancer sites studied. The findings supported hypotheses based on the results of animal experiments showing that different kinds of fatty acids have different tumor-promoting capabilities. o two Academic PESS, IIIC.


There has been little exploration of the statistical relationships of different types of dietary fats with human cancers, although a number of literature reports suggest an association between meat and other food items high in saturated fat (SF) and the incidence of cancer of the breast and colon (l-4). Prospective studies of the morbidity and mortality of individuals with different baseline levels of serum cholesterol (5-7), and an analysis of cancer incidence in the Mormon population (8) have raised some concerns about the possible associations of polyunsaturated fat (PUF) with cancer incidence as well. The body of evidence from experimental studies in animals, recently reviewed elsewhere (9, lo), supports a tumor-promoting role for both PUF and SF in various mammary and colon tumor model systems, especially when the essential i Present address: Department of Nutrition, Box 7400 MacGavran-Greenberg Hall, University of North Carolina, Chapel Hill, NC 27514. ’ To whom reprint requests should be addressed. 242 0091-7435/90$3.00 Copyright 8 19!20 by Academic Press, Inc. All rights of reproduction in any form reserved.





fatty acid requirements are met. However, the type-specific effects of PUF and SF have not been resolved. Their differential effects on tumor promotion at low levels of essential fatty acids suggest a mechanism other than their caloric contribution, since at these levels PUF is a better promoter than SF (11-16) despite similar caloric values and efftciency of utilization (17). Studies using monounsaturated fat (MUF) report that it is not an effective promoter of chemically induced tumors in rodents (11, 15). Some recent data suggest a difference in the cancer promotional effects of the two subclasses of PUF. w-3 polyunsaturated fatty acids (n-3 PUF), which come primarily from marine lipids, can reportedly inhibit rodent mammary (8, 11, B-20), colon (21), and prostate (22) tumorigenesis and tumor growth, while w-6 polyunsaturated fatty acids (n-6 PUF), which are the primary subclass of fatty acids in most vegetable oils, consistently promote tumors in the same studies. The putative anticancer effect of n-3 PUF is indirectly supported by epidemiologic data on Greenland Eskimos, who consume large amounts of n-3 PUF (23) and have relatively low reported incidence rates of cancer (24). In addition, Kaizer et al. (3) recently reported that per capita percentage calories from fish, estimated from international food disappearance data for 32 countries, had a significant inverse correlation with breast cancer incidence. Since the largest pieces of evidence for the relationship between some human cancers and total fat intake have come from analyses of the results of natural experiments between countries (25-30) and studies of migrants from countries with low and countries with high rates of breast cancer (3 l-37), these data are the best available source of information about more specific associations (38). So far, only a few studies have attempted to look at the relationship between types of fat and any single cancer site using such data. Carroll (26), and more recently Wynder (30), correlated age-adjusted breast cancer mortality rates from various countries with per capita disappearance of animal and vegetable fat and found that breast cancer mortality was strongly associated with animal fat, but not strongly associated with vegetable fat. Using multiple regression analysis to control for potentially confounding variables and careful delineation of outcome data, Armstrong and Doll observed a strong correlation between animal fat and breast and colon cancer incidence and mortality (25). In support of international differences, Kolonel et al. correlated individual fat consumption with ethnic patterns of breast and prostate cancer incidence in Hawaii and found significant associations between breast cancer and saturated and unsaturated fats, and between prostate cancer and saturated fats (29). The present study is a detailed correlation analysis of international data to explore the relationships among PUF (total and fish n-3 and n-6), SF, and MUF and the incidence of cancers of the breast, colon, and prostate, for which a positive association was expected, and lung and cervix, with no anticipated strong fat association. The most precise measurements of total and component fat disappearance that could be derived, and the most reliable available measurements of national cancer incidence, were used in a multiple regression analysis to calculate partial correlation coefftcients and slopes for different types of fat and the incidence of cancer at the five anatomic sites. An exhaustive analysis of the relative amounts of fatty acids in reported foodstuffs was performed to develop






the exposure variables. The methods used by Armstrong and Doll (25) were adopted to optimize the reliability of the outcome variables. All analyses were standardizedfor estimated total caloric intakes, as suggestedby Wlett (39). Also, the role of dietary fiber as an interacting variable with fat in breast and colon cancer was evaluated. The relationship between dietary fiber consumption and colon cancer has been reviewed elsewhere (40). Interest in a possible protective role for fiber, through its alteration of the route of excretion of estrogens, in breast carcinogenesisstemsfrom the studies of Gorbach and Goldin et al. (41,42). These international dietary and cancer incidence data sets were used to test three hypotheses: (a) total fat is positively and significantly associated with breast and colon cancer incidence among women, and cancers of the colon and prostate among men. It is not significantly associated with the incidence of lung and cervical cancers among women and lung cancer among men; (b) PUF and SF are associated with breast and colon cancers among women, and with colon and prostate cancers among men; (c) n-6 PUF is positively, and fish n-3 PUF is negatively, associatedwith the breast and colon cancer incidence among women, and colon and prostate cancers among men. METHODS

Average cancer incidence data for breast (female), cervix (female), prostate (male), colon (male and female), and lung (male and female) for the years 19731977were taken from Cancer Incidence in Five Continents (43). Criteria described by Armstrong and Doll (25) were used to select countries and incidence parameters to be included in the analysis (Table 1). Incidence rates, truncated to the age range of 35-64 years and age-standardized against the world population, were used for the following reasons: (a) at these sites, and at these ages, cancer is common enough to produce stable rates in relatively small populations; (b) in the truncated range, the relationship between incidence and age is similar in virtually all countries; (c) inaccuracies are minimized by excluding data from the oldest age groups, in which cancer registration is likely to be very incomplete; and (d) this age range is the most relevant for the chosen cancer sites, which have their maximum incidence in later adult life (25). For countries in which data were available from two or more cancer registries, a mean truncated rate was calculated and weighted for the population in the 35 to 64-year age group in each registration area. Estimates of per capita disappearance of total fat, PUF, SF, MUF, fish n-3 PUF, n-6 PUF, and total calories for the 20 countries included in the analyses (Table 2) were calculated or taken directly from Food Balance Sheets published by the United Nations Food and Agricultural Organization (44). The Food Balance Sheets, averaged for the period 1975-1977,were concurrent with the incidence data. Component fats were calculated from these data by (a) extracting the total per capita grams of fat supplied by each of 68 fat-containing food items listed in the Food Balance Sheetsfor each country; (b) multiplying the per capita grams of total fat supplied by each food item for each country by the published (17, 27) component proportion of total fat (PUF, SF, MUF, and fish n-3 PUF) in each food item; and (c) summing the grams of each component fat from each food item in









Australia Ihlada Denmark Finland France Hong Kong Hwwv Israel IdY Japan New Zealand Norway Romania Spain Sweden Switzerland United Kingdom United States West Germany Yugoslavia

118.3 143.1 128.6 87.9 128.9 68.7 53.8 123.0 129.6 35.2 139.0 109.0 66.2 86.4 117.1 145.5 125.1 163.2 127.5 80.4

Male colon

Female colon


29.6 28.5 21.7 10.9 28.1 20.8 15.6 15.7 23.5 11.6 35.8 18.4 8.4

26.9 32.0 24.6 12.0 20.2 18.2 14.0 16.7 22.3 9.9 41.5 20.8 4.3 9.4 20.4 17.9 21.7 31.8 21.0 9.2

22.5 13.6 15.9 13.7 3.7 6.6 10.9 12.2 2.1 16.3 23.5 4.7 8.6 27.3 16.9 10.6 33.7 20.9 10.2


19.2 22.2 20.2 33.4 21.1 12.7


Male lung

Female lung

75.7 82.4 76.9 120.9 94.1 92.2 64.0 42.2 125.4 32.9 87.3 42.8 58.5 41.7 33.2 68.1 115.9 99.4 %.4 90.0

16.2 21.8 23.2 9.8 5.9 38.6 10.5 14.3 10.5 11.1 28.4 10.2 9.1 5.0 9.4 14.5 30.8 39.5 13.4 10.8


24.6 31.1 55.9 19.0 42.1 72.8 30.2 10.9 27.5 43.0 28.9 44.5 70.2 12.7 28.5 32.7 27.0 24.2 47.0 37.8

each country, to get the total grams of PUF, SF, MUF, and fish n-3 PUF for each of 20 countries (see Table 2). An estimate of n-6 PUF was obtained by subtracting fish n-3 PUF from total PUF. Fish n-3 PUF was used becausethe databasefor n-3 in foods other than fish is substantially incomplete. The sum of the PUF, SF, and MUF, calculated from 68 food items, as described above, accounted for approximately 90% of the total fat supply published for each country in the Food Balance Sheets. Total per capita calorie counts were taken directly from the Food Balance Sheets. Estimated per capita dietary and crude fiber consumption, calculated from the same source as the fat data, was extracted from Bright-See and McKeown-Eyssen (45). Multiple regression analysis (46) was used to assessthe associations between the cancer incidence rates at each site for the 20 countries under study and the estimated intake of total PUF, SF, MUF, fish n-3 PUF, and n-6 PUF, as well as the relative intakes of PUF:SF and fish n-3:n-6 PUF. All analyses were standardized for estimated total calorie intakes. A separate analysis evaluating fiber as an interactive variable was also performed. RESULTS

The independent variables entered into the regression analysis were total fat, SF, PUF, n-6 PUF, fish n-3 PUF, MUF, total calories, and dietary and crude fiber. Table 3 lists correlation coefftcients for each pair of fat consumption variables








Australia Canada Denmark Finland France Hong Kong Hungary Israel Italy Japan New Zealand Norway Romania Spain Sweden Switzerland United Kingdom United States W. Germany Yugoslavia

Total Fat 0

PUF km)

SF (gm)

MUF f&n)

n-3 PUF (an)

n-6 PUF km)

Total calories

130 149 161 135 148 99 133 112 122 72 148 146 86 127 140 155 140 164 155 93

17.1 21.1 22.1 16.5 24.8 19.9 20.6 32.5 23.4 19.7 14.7 29.0 22.4 30.2 19.6 27.0 20.5 33.4 25.9 23.6

56.1 56.6 62.9 59.0 55.9 26.6 47.8 32.2 36.5 19.0 68.8 53.9 28.3 33.8 54.9 60.0 57.2 56.1 55.6 30.4

40.7 59.4 64.8 49.2 54.8 41.6 53.7 36.3 52.3 24.2 44.5 52.8 27.8 51.6 54.7 56.2 50.8 61.4 61.7 31.8

0.04 0.04 0.44 0.37 0.28 0.50 0.15 0.15 0.12 1.5 0.07 0.40 0.13 0.38 0.50 0.12 0.13 0.10 0.14 0.06

17.1 21.1 21.7 16.1 24.5 19.4 20.4 32.3 23.3 18.2 14.6 28.6 22.3 29.8 19.1 26.9 20.4 33.3 24.8 23.5

3,400 3,345 3,430 3,190 3,458 2,672 3,494 3,148 3,469 2,845 3,422 3,124 3,372 3,214 3,168 3,385 3,311 3,539 3,361 3,469

Note. PUF, polyunsaturated PUF; n-6 PUF, o-6 PUF.

fat; SF, saturated fat; MUF,


fat; n-3 PUF, fish 0-3

and Tables 4 and 5 display the partial correlations and slopes derived from multiple regression analysis of breast, colon, prostate, lung, and cervical cancer incidence on those fat-consumption variables. Only the results for lung and colon cancer incidence among women are shown in these tables3 and discussed further, as the patterns of results from lung and colon cancer among men were virtually the 3 Data for males are shown in Table 1. TABLE




Simple correlation

Total fat, SF Total fat, PUF Total fat, MUF PUF, SF PUF, MUF SF, MUF Fish n-3 PUF, n-6 PUF Note. PUF, polyunsaturated PUF; n-6 PUF, w-6 PUF.

fat; SF, saturated fat; MUF,


0.91 0.17 0.92 -0.18 0.22 0.73 -0.21 monounsaturated

fat; n-3 PUF, fish o-3



PUF (SF, MUF, Cab) Cancersite



Femalebreast 0.51(0.036”) 2.1 Femalecolon -0.01 -0.01 Prostate Female lung Cervix

0.46 (0.063) 0.02 -0.26

0.64 0.06 -0.97

(PUF, h%F, Cab) S PCC 1.8 0.58 (0.015) 0.44 0.47 (0.056) 0.43 0.55 (0.022) 0.24 0.19 -0.46 -0.23


MUF (PUF, SF, Gals) PCC -0.01 0.004 0.02 0.04 0.15




-0.03 0.004 0.01 0.05 0.35

0.72 (.00046) 0.62 (.0048) 0.69 (~3011) 0.34 -0.17

0.95 0.24 0.23 0.16 -0.12

Note. PCC, partial correlationcoefficientbetweencancerincidencerate for the tabulatedsite and the tabulated dietary component(adjustedfor the dietary component(s) in parentheses). S, slope of the least-squares relationship between the cancer incidence rate and tbe dietary component (adjusted for the dietary component(s) in parentheses).

Unit = cases/1OO,OOO/g. PUF, polyunsaturatedfat; SF, saturatedfat; MUF, monounsaturated fat; TF, total fat; Cak, calories. a P value of test of PCC = 0; only P values below 0.10 are shown.

same as those among women. Because the intakes of component fats are correlated with each other and with total fat consumption, all regression analyses were controlled for intakes of all other component fats as well as for total calories. Under this control, there were strong positive associations between cancer incidence of the breast, colon, and prostate and total fat consumption and very low probabilities that these were chance statistical findings. Associations between SF and cancer site incidence were fully in line with predictions. Saturated fat was strongly associated with the incidence of breast, colon, and prostate cancers. Polyunsaturated fat had the predicted strong positive associations with breast and prostate cancers but not with colon cancer. Neither total fat nor SF nor PUF had a strong positive association with lung or cervical cancer. Fish n-3 PUF had a nonsignificant negative association with breast, colon, TABLE 5 PARTIAL CORRELATION COEFFICIENTS BETWEEN AGE-STANDARDIZED CANCERINCIDENCERATES AND POLYUNSATURATED FAT COMPONENTS

fish n-3 PUF (MUF, SF, n6 PUF, Cals)

n-6 PUF (MUF, SF, fish n3 PUF, Cals)

Cancer site





Female breast Female colon Prostate Female lung Cervix

-0.28 -0.15 0.04 -0.29

-23.4 -4.3 0.9 -12.1

0.50 (0.049”) -0.03 0.46 (0.074”)

2.5 - 0.05 0.6

- 0.03


- 0.02 -0.26



Note. PCC, partial correlation coefhcient between cancer incidence rate for the tabulated site and the tabulated dietary component (adjusted for the dietary component(s) in parentheses). S, slope of the least-squares relationship between the cancer incidence rate and the dietary component (adjusted for the dietary component in parentheses). Unit = cases/lOO,OOO/g. PUF, polyunsaturated fat; SF, saturated fat; MUF, monounsaturated fat; Gals, calories; n-3 PUF, fish o-3 PUF; n-6 PUF, all other PUF. n P value of test of PCC = 0; only P values below 0.10 are shown.






lung, and cervical cancers. The estimate of n-6 PUF (total PUF minus fish n-3 PUF) gave virtually the same results as total PUF. The effect of fiber on these associations was examined. Adjustment for dietary fiber reduced the correlation between total fat and colon cancer incidence from 0.62 to 0.31 and reduced the correlation between SF and colon cancer from 0.47 to 0.31. Adjustment for crude fiber resulted in similar reductions. In all other analyses, the effect of dietary or crude fiber on the measured degree of association was negligible. As expected, MUF was not associated with cancer at any studied site. Total calories had no association with cancer incidence at any site when the analysis was controlled for total fat consumption. DISCUSSION A majority of previous ecologic studies of dietary fat and cancer have used relatively crude measurements of PUF and SF intakes (vegetal and animal fat disappearance), mortality rather than incidence rates, and simple regression analysis. The results of the more stringent analysis presented here tend to reinforce previous findings. One major constraint of this ecologic approach is that dietary fat consumption (disappearance) increases with social and economic development and may simply be a marker of other environmental changes including improved cancer detection. However, analyses which incorporated all available measurements of economic development did not reduce the strength of the dietary fatbreast cancer incidence association (10). Furthermore, the different relationships of fat consumption with different cancers and of site-specific incidence with different fats suggest that some components of the relationship go beyond that of an index of social development per se. The predicted hypotheses were derived from a review of the animal literature on promotion of mammary and colon tumors. The consistency of the results of the multiple regression analyses with those hypotheses provide some degree of confidence in the specificity of the associations. Because these analyses were restricted to 20 countries judged to have reliable nationwide cancer incidence statistics, an analysis was carried out to see if data from one or a small group of these countries could have exerted an undue influence on the results. No single country had a consistently large impact on the correlations. Data from population-based dietary surveys in the United States (7) and Australia (47) confirm that individual citizens in those countries have diets that are high in fat. There are, unfortunately, no representative records of dietary intakes of individuals living in each country included in these analyses. In the data analyzed, the associations between total fat consumption and the incidence of breast, colon, and prostate cancer appear to be independent of total calorie intake and of the ratio of polyunsaturated to saturated fat intake. The documentation of this independent effect is important since high levels of fat consumption are found in countries with high overall caloric intakes, and high P: S ratios are usually found in countries where overall caloric and fat intakes are low. In addition, one hypothesis for the role of dietary fat in the promotion of mammary cancer and other fat-related cancers suggests an indirect mechanism





through a caloric-effect rather than a specific fat effect involving hormones or other host factors (29). Thus, an important finding of this study is that our data contradict this hypothesis. The results of the controlled analysis of total fat consumption somewhat underestimates the strength of the association between the consumption of tumorpromoting dietary fats and cancer becauseat least 30% of dietary fat is consumed in the relatively non-tumor-promoting monounsaturated form. When tumorpromoting fats are analyzed independently, saturated fat has a consistent, and polyunsaturated fat an inconsistent, pattern of association with the three fatrelated cancer sites. The strong association of colon cancer with saturated, but not with polyunsaturated, fat was unexpected. If anything, reports of an association of colorectal cancer with low levels of serum cholesterol raised concerns about intakes of diets with high P:S ratios (48, 49). This finding was in agreement with Carroll (26) and McKeown-Eyssen (50), both of whom used the less specific fat variables (vegetableand animal fats) as estimatesof PUF and SF and colon cancer mortality rates rather than incidence. However, in contrast to Carroll’s study (26), which included mortality data from countries that did not meet Armstrong and Doll’s selection criteria and used only simple correlation analysis, we did find a significant association between PUF and breast cancer. The discrepancy is possibly due to our use of more reliable outcome measurements, our control of potentially confounding variables by multiple regression analysis, and our ability to separate animal and vegetable fat into PUF, SF, and MUF. The difference between the fat association of the hormone-related breast and prostate cancers and colon cancer is worth further consideration. The overriding importance of total fat consumption in these analyses does not therefore rule out some underlying interplay of component dietary fats. The attempt to analyze the independent associations of n-6 and fish n-3 PUF was constrained by the small and relatively invariable amounts of fish n-3 PUF consumed. The results obtained did, however, show a nonsignificant but consistent negative association between fish n-3 PUF and the cancer sites studied. The strongest negative correlation was between fish n-3 PUF and breast cancer, in accord with the report from Kaizer et al. (3) of a negative association between per capita percentage calories from fish and international breast cancer incidence rates. Non-fish PUF, an estimate for n-6 PUF, had essentially the same correlations as total PUF. Thus our findings from human data provide further support, albeit weak, for the suggestion from animal experiments that n-3 fatty acids found in certain fish may protect against mammary and possibly other types of cancers. Confirmation of a differential effect of n-3 and n-6 PUF on human cancer incidence will require analytic epidemiological studies or possibly human trials. Given that animal studies also suggest an inverse relationship between n-3 PUF and progression of certain tumors (8, 19), a relationship with mortality rather than incidence data should also be investigated. While these analyses of international data cannot be used to draw conclusions about the role of fat in cancer causation, the results do strengthen the hypothesis that positive associations between fat consumption and cancer incidence are sitespecific and that different fats have different cancer incidence associations. Not




all studies agree with our findings. Reports by Jensen et al. (51), McMichael et al. (52), and Enstrom (53) question a positive association between animal fat and colon cancer. In addition, there are contradictory results in the animal experimentation literature, recently reviewed by Rogers and Nauss (54), on the role of dietary fat in colon carcinogenesis. Our study indicates that fiber is an important confounding variable in the association between dietary fat and colon cancer, which is not surprising in light of the evidence of a protective effect of fiber on colon cancer (40). Also, although the totality of studies in the literature is suggestive of a causal relationship between fat intake and breast cancer (30), the largest analytic epidemiological study to date assessing this relationship, the Nurses Health Study, reported no association between dietary fat and breast cancer (55). However, whether the dietary assessment instrument used in that study could detect differences across the relatively narrow range of dietary fat intakes of the cohort studied is questionable (10). If indeed different fatty acids are associated with cancer incidence at different sites, fatty acids could be presumed to have both general and specific influences. The general influence could be through the provision of energy; the more specific through one or more of a number of plausible physiological mechanisms, possibly differing in cancers at different sites (30). CONCLUSION

There is little likelihood that definitive analytic epidemiologic studies of dietary fat and cancer of individual sites will be forthcoming. A recent estimate called for at least 4,ooO cases and 4,000 controls before the risk suggested by international variations in breast cancer incidence could be detected in a case-control study conducted in the United States (2). A very critical question is the extent to which we can rely on evidence from international differences in human populations and animal data as the justification for human public health experiments. The data used in the present study are subject to all the inconsistencies of gross national figures. Given the relatively crude nature of the food disappearance data in particular, the degree of agreement between the initial hypotheses and the observed results using this human data lends evidence to the findings in experimental animals and thus supports the proposition that there is an underlying biological relationship between relative fat intake and variations in human cancer. The importance of the relationship suggested by these data to the mass prevention of human cancers requires a more definitive investigation. REFERENCES 1. Dales LG, Friedman GD, Ury HK, Grossman S, Williams SR. A case-control study of relationships of diet and other traits to colorectal cancer in American blacks. Am .I Epidemiol 1978; 109~132-144. 2. Goodwin PJ, Boyd NF. Critical appraisal of the evidence that dietary fat intake is related to breast cancer risk in humans. .I Nutl Cancer Znst 1987; 79473-485. 3. Kaizer L, Boyd NF, Kriukov V, Tritchler D. Fish consumption and breast cancer risk: An ecological study. Nutr Cancer 1989; 12~61-68.





4. Lubin JH, Burns PE, Blot WJ, Ziegler RG, Lees AW, Fraumeni JF. Dietary factors and breast cancer risk. Znt J Cancer 1981; 28:685-9. 5. Bjelke E. Colon cancer and blood cholesterol. Lancet 1974; 1:1116-l 117. 6. Kritchvesky D, Klurlield DM. Caloric effects in experimental mammary tumorigenesis. Am J C/in Nutr 1987; 45~236-242. 7. National Center for Health Statistics. Dietary intake source data: United States, 1976-1980. Vital and Health Statistics Series 11, No. 231. DHHS Publ. No. (PHS) 83-1681. 8. Karmali RA, Marsh J, Fuchs C. Effect of omega-3 fatty acids on growth of a rat mammary tumor. J Natl. Cancer Znst 1984; 73:457-461. 9. Ip C, Bitt DF, Rogers AE, Mettlin C. Dietary fat and cancer. Prog Clin Biol Res 1986; 2221-885. 10. Prentice RL, Kakar F, Hursting S, Sheppard L, Klein R, Kushi L. Aspects of the rationale for the Women’s Health Trial. J Nat1 Cancer Znst 1988; m802-814. 11. Braden LM, Carroll KK. Dietary polyunsaturated fat in relation to mammary carcinogenesis in rats. Lipids 1986; 21:285-288. 12. Broitman SA, Vitale JJ, Vavrousek-Jakuba E, Gottlieb LS. Polyunsaturated fat, cholesterol and large bowel tumorigenesis. Cancer 1977; 40:2455-2463. 13. Chan PC, Head JF, Cohen LA, Wynder EL. Influence of dietary fat on the induction of mammary tumors by N-nitrosomethylurea: Associated hormone changes and differences between Sprague-Dawley and F-344 rats. J Nat1 Cancer Znst 1977; 59:1279-1283. 14. Clinton SK, Mulloy AL, Visek WJ. Effects of dietary lipid saturation on prolactin secretion, carcinogen metabolism and mammary carcinogenesis in rats. J Nutr 1984; 114:1630-1639. 15. Cohen LA, Thompson DO, Maeura Y, Choi K, Blank ME, Rose DP. Dietary fat and mammary cancer. I. Promoting effects of different dietary fats on N-nitrosomethylurea-induced rat mammary tumorigenesis. J Nat1 Cancer Znst 1986; 77~3342. 16. Rogers AE, Wetsel WC. Mammary carcinogenesis in rats fed different amounts and types of fat. Cancer Res 1981; 41:3735-3737. 17. Donato K. Efficiency and utilization of various energy sources for growth. Am J Clin Nutr 1987; 45:164-167. 18. Gabor H, Abraham S. Effect of dietary menhaden oil on tumor cell loss and the accumulation of mass of a transplantable mammary adenocarcinoma in BALB/c mice. J Nat1 Cancer Znst 1986; 76:1223-1229. 19. Ip C, Ip MM. Relevance of trans fatty acids and fish oil in animal tumorigenesis. Prog Clin Biol Res 1986; 222:283-294. 20. Jurkowski JJ, Cave WT. Dietary effects of menhaden oil on the growth and membrane lipid composition of rat mammary tumors. J Nat1 Cancer Znst 1985; 741145-l 150. 21. Reddy BS, Maruyama H. Effect of dietary fish oil on azoxymethane-induced colon carcinogenesis in male F-344 rats. Cancer Res 1986; 46:3367-3370. 22. Karmali RA, Reichel P, Cohen LA. Dietary effects of omega-3 fatty acids on the growth of the DU-145 prostatic tumor. Fed Proc 1986; 45, Abstract 477. 23. Bang HO, Dyerberg J, Hjorne N. The composition of food consumed by Greenland Eskimos. Acta Med Stand 1976; 200:69-73. 24. Nielsen NH, Hansen JP. Breast cancer in Greenland-Selected epidemiological, clinical and histological features. J Cancer Res Clin Oncol 1980; 98:287-299. 25. Armstrong B, Doll R. Environmental factors and cancer incidence and mortality in different countries, with special reference to dietary practice. Znt J Cancer 1975; 15:617431. 26. Carroll KK. Lipids and carcinogenesis. J Environ Pathol Toxic01 1980; 3:253-271. 27. Gray GE, Pike ME, Henderson BE. Breast cancer incidence and mortality rates in different countries in relation to known risk factors and dietary practices. Br J Cancer 1979; 391-7. 28. Hems G. The contributions of diet and child bearing to breast cancer rates. Br J Cancer 1978; 37974-982. 29. Kolonel LN, Hankin JH, Lee J, Chu SY, Nomura AMY, Hinds MW. Nutrient intakes in relation to cancer incidence in Hawaii. Br J Cancer 1981; 44:332-339. 30. Wynder ER. Amount and type of fat/fiber in nutritional carcinogenesis. Prev Med 1987; 16:451459.




31. Adelstein AM, Staszewski J, Muir CS. Cancer mortality in 1970-1972among Polish born migrants in England and Wales. Br J Cancer 1979; ti464-475. 32. Buell P. Changing incidence of breast cancer in Japanese-American women. J Nat1 Cancer [nrr 1973; 51:1479-1483. 33. Haenszel W. Cancer mortality among foreign born in the United States. J Nat1 Cancer Inst l%l; X37-132. 34. King H, Locke FB. Cancer mortality among Chinese in the United States. J Nat1 Cancer Znst 1980; 65:1141-l 148. 35. Margetts BM, Hopkins SM, Birms CW, Miller MR, Armstrong BK. Nutrient intakes in Italia.n migrants and Australians in Perth. Food Nutr 1981; 38:7-10. 36. Staszewski J, Haenszel W. Cancer mortality among Chinese in the United States. J Nat1 Cancer Znst 1%5; 35:292-297. 37. Staszewski J, McCall MG, Stenhouse NS. Cancer mortality in 1962-1966among Polish migrants to Australia. Br J Cancer 1971; 25:599-610. 38. Pike MC, Ross RK. Breast cancer. Br Med Bull 1984; 40:351-354. 39. Willett W. Implications of total energy intake for epidemiologic studies of breast and large bowel cancer. Am .I Clin Nutr 1987; 49354-360. 40. Khutield DM, Kritchvesky D. Dietary fiber and human cancer: Critique of the literature. Adv &p Med Biol 1986; 206:119-135. 41. Goldm BR, Adlercreutz H, Gorbach SL, Woods MM, Cordon T, Bohn E, Gershoff SN. The relationship between estrogen levels and diets of Caucasian American and Oriental immigrant women. Am J Clin Nutr 1986; 44:945-953. 42. Gorbach SL, Goldin BR. Diet and the excretion and enterohepatic cycling of estrogens. Prev Med 1987; 16525-531. 43. Waterhouse J, Shanmugaratham K, Muir C, Powell J, Eds. Cancer Incidence in Five Continents, Volume IV. Lyon: WHO International Agency for Research on Cancer, 1982. 44. Food and Agricultural Organization of the United Nations. Food Balance Sheets: 1975-1977 Average. Rome, 1980. 45. Bright-See E, McKeown-Eyssen GE. Estimates of per capita crude and dietary fiber supply in 38 countries. Am J Clin Nutr 1984; 39821-829. 46. Snedechor G, Co&ran W. Statistical Methods. Ames: Iowa State Univ. Press. 47. McMichael AJ, Giles GG. Cancer in migrants to Australia: Extending the descriptive epidemiological data. Cancer Res 1988; 48:751-756. 48. Committee of Principal Investigators. WHO cooperative trial on primary prevention of ischaemic heart disease using clotibrate to lower serum cholesterol: mortality follow-up. Lnncet 1980; 2379-385. 49. Nydegger UE, Butler RE. Serum lipoprotein levels in patients with cancer. Cancer Res 1972;

321756-1760. 50. McKeown-Eyssen GE, Bright-See E. Dietary factors in colon cancer: International relationships. Nutr Cancer 1989; 6:160-170. 51. Jensen OM, MacClennan R, Wahrendorf J. Diet, bowel function, fecal characteristics and large bowel cancer in Denmark and Finland. Nutr Cancer 1982; 4:5-19. 52. McMichael AJ, Potter JD, Hetzel BS. Time trends in colorectal cancer mortality in relation to food and alcohol consumption in the United States, United Kingdom, Australia and New Zealand. Znt J Epidemiol 1979; 8295-303. 53. Enstrom JE. Colorectal cancer and consumption of beef fat. Br J Cancer 1975; 32432-439. 54. Rogers AE, Nauss KM. Rodent models for carcinoma of the colon. Digest Dis Sci 1983; 30:89S102s. 55. Willett WC, Stampfer UT, Colditz GA, Rosner BA, Hennekens CH, Zpeizer FF. Dietary fat and the risk of breast cancer. N Engl J Med 1987; 316:22-28. 56. Haenszel W, Berg JW, Segi M, Kurihara M, Locke FB. Large bowel cancer in Hawaiian Japanese. J Nat1 Cancer Znst 1973; 51:1765-1779. 57. Hepburn FN, Exler J, We&much JL. Provisional tables on the content of omega-3 fatty acids and other fat components of selected food. J Am Diet Assoc 1986; 86:788-793.





58. Lyon JL, Gardner JW, West DW. Cancer incidence in Mormons and non-Mormons in Utah during 1%7-1975. .ZNat1 Cancer Ins? 1989; 65:105&1061. 59. Phillips RL. Role of lifestyle and dietary habits in risk of cancer among Seventh-Day Adventists. Cancer Res 1975; 35:3513-3522. 60. Reeves JB, Weihrauch JL. Composition of Foods, Agriculture Handbook Number 8 (1-12). Washington, DC: U.S. Department of Agriculture, 1979. 61. Rose GH, Blackbum H, Keys A, Taylor HL, Kannel WB, Paul 0, Reid DD, Stamler J. Colon cancer and blood cholesterol. Lancet 1974; 1:181-183.