Gut: first published as 10.1136/gut.10.12.1031 on 1 December 1969. Downloaded from

Gut, 1969, 10, 1031-1034

Gastrointestinal cancer and nutrition

0. GREGOR1, R. TOMAN, AND F. PRUSOVA From the First Medical Department, University Hospital, and the Research Institute of Trade, Prague, Czechoslovakia

SUMMARY The hypothesis upon which this study was based is that there is a relationship between mortality from gastrointestinal cancer and living standards. On this basis we found signi- ficant correlations between the intake of animal proteins and the mortality rates for gastric and in- testinal cancer. The negative correlation coefficient (r = 0.85) is an expression of the inverse relationship between gastric and intestinal cancer mortality rates. This inverse relationship is also expressed as the correlation between the food intake, expressed by the intake of animal protein, and the respective mortality rates. The higher the food intake, the lower the gastric cancer mortality rate but the higher the intestinal cancer mortality rate. We do not claim that this relationship discovered by correlation analysis is a causal one. On the basis of this study it cannot therefore be said that food intake has a direct effect on the development ofgastrointestinal cancer. In this respect our findings can only be a signal for further studies. Secondly no time lag has been proved between food intake and the mortality rate for intestinal cancer. The findings relating to gastric cancer do not contradict the hypothesis of a time lag.

The aetiology of gastrointestinal cancer is not various nutrients, the closest correlation being that known. Its pathogenesis seems to be multifactorial with the intake of animal proteins (r = 0.77) (Acheson and Doll, 1964; Jones, 1966; Gregor, (Dewhurst, Coppock, Yates, and their associates, 1965; Gregor, Toman, Pastorova, and Drnkova, 1961). These findings formed the basis for our 1966; Gregor, Toman, Prusova, Drnkovai, and hypothesis that there exists a relationship between http://gut.bmj.com/ Pastorova, 1969; Hirayama, 1963; Saxen and the mortality from gastrointestinal cancer and the Hakama, 1967; Schwartz, Lassere, Flament, and standard of living. Our second problem has been to Lellough, 1966; Segi and Kurihara, 1966; Wynder, determine whether a time lag between the intake of Kmet, Dungal, and Segi, 1963; Wynder, Graham, nutrients and mortality from gastrointestinal cancer and Eisenberg, 1966; Wynder and Shigematsu, can be proved. 1967), one factor being the standard of living. Mortality from gastric cancer in the different social MATERIALS on September 23, 2021 by guest. Protected copyright. classes has been studied by many authors and correla- tions with income level have been found (Backett, We have taken as the basis for our study data on stand- 1967; Clemmesen, 1965; Doll, 1956; Haenszel, ardized mortality rates for cancer of the and 1958; Higginson, 1966; Mider, 1960): the higher the the intestine (Segi and Kurihara, 1966) and data on the income, the lower the mortality. Earlier studies have intake of nutrients (Dewhurst et al, 1961; Food and been concerned with this relationship within Agriculture Organization, 1948-65) in 28 countries (Table I), and the comparison of the intake of nutrients countries but we have tried to study the problem in the USA and in Czechoslovakia from the respective on an international scale. statistical yearbooks of these countries (United States, We have found a correlation between the rate of 1964; Czechoslovakia, 1966). private expenditure per head, which we have calcu- lated from the gross national product per head minus expenditure on consumer goods minus the METHODS gross growth of domestic capital, as the general We used correlation analyses as the fundamental method indicator of the standard of living, and the intake of study: the gastric and intestinal cancer mortality rates 'Please address requests for reprints to 0. Gregor, Vlagski 36, for 1962-63 were correlated with the intake of animal Prague 1. piotein at two different periods, namely, 1962-63 and 1031 Gut: first published as 10.1136/gut.10.12.1031 on 1 December 1969. Downloaded from

1032 0. Gregor, R. Toman, and F. Prusova 1947-48, and the correlation coefficients tested against the critical values laid down by Fisher and Yates (1949). Calories TABLE I Proteins BASIC DATA ON STANDARDIZED MORTALITY RATES (MALE) Animal proteins FOR STOMACH AND INTESTINAL CANCER AND THE DAILY Fats INTAKE OF ANIMAL PROTEIN IN 28 COUNTRIES. Country Standard Mortality Daily Intake of Carbohydrates Rate Animal Protein Vitamin A Stomach Intestines 1947-48 1962-63 Austria 43.04 9.59 21-1 47.5 Vitamin B, Belgium- Luxembourg 29-46 12-99 33-1 45.9 Vitamin B2 Denmark 23-93 13.91 59-1 57.9 Finland 44-80 5S20 34-6 54.7 Vitamin C France 23-03 11-34 29-7 43.0 0l 25 so 75 100 Federal Republic of Germany 38-41 9 15 21-1 49-2 %/ of U.S. intake Greece 15-17 4.14 15-2 30-6 Ireland 25-68 11-86 46-7 54-2 Italy 34-22 7.97 15-2 29-8 FIG. 1. Comparison between intake of food in Czecho- Netherlands 29.50 10.53 34.4 46-1 the constituents ofthe Norway 27-67 7-89 47.9 48-8 slovakia and USA broken down into Portugal 32-60 6-65 18-0 27-3 total calories consumed. Spain 30.37 6-15 18.2 23-4 Sweden 24-83 9-78 62-6 54.3 Switzerland 27-81 10-53 43-8 51-3 findings do not contradict the assumption of a UAR 24-90 12-43 47-6 53.4 time lag existing between the intake of food and the Yugoslavia 21-72 2-16 11-2 25-5 mortality from gastric cancer. The correlation Poland 39.79 2-95 17-3 40-0 Czechoslovakia 43.59 6-49 28-3 43 0 coefficients between the gastric cancer mortality Hungary 46-40 6-26 15-9 37.0 rates and the intake of animal proteins are negative, Canada 18-69 13-88 59-2 63-8 USA 14-87 12-25 61-4 64-2 whereas those between the intestinal cancer mortality Australia 17-06 12-68 59.9 59.6 rates and the intake of animal proteins are positive. New Zealand 17-90 12-45 72-5 74-8 Chile 64-63 2-73 24-7 29-2 The inverse relationship existing between the Colombia 21.20 2-24 22-5 20-0 mortality rates for gastric and intestinal cancer Venezuela 33.00 4.10 19-5 25-3 (r = - 085) (Segi and Kurihara, 1966) means that http://gut.bmj.com/ Japan 67-96 2-97 9.7 16-9 the higher the food intake, the lower the mortality The correlation coefficients lying above the critical rate from gastric cancer and the higher the food in- value at the 5% level were considered as significant, take (expressed as animal protein), the higher the those lying below the critical value at the 1 % level as not mortality rate from intestinal cancer. significant. The correlation coefficients lying between the The results ofthe correlation analyses are shown in two critical values were considered doubtful. Table II and Figures 2a and b and Figures 3a and b. on September 23, 2021 by guest. Protected copyright. RESULTS TABLE II CORRELATION COEFFICIENTS BETWEEN THE INTAKE OF A comparison between food intake in the USA and ANIMAL PROTEIN AND THE MORTALITY RATES FOR GASTRIC in Czechoslovakia is shown in Figure 1. There were AND INTESTINAL CANCER practically no differences in the total intake of Correlation Coefficient calories but differences were most significant for animal protein, fats, carbohydrates, and vitamins Period of Study Gastric Cancer Intestinal Cancer A5B11 2, and C (Fig. 1). 1962/63 -0442 (?) +0-810 (s) The correlation coefficients between the mortality 1947/48 -0-530 (s) +0-760 (s) rates for intestinal cancer and the intake of animal 's = significant (above + 0.4869) proteins are significant. There are no significant ? = doubtful + (0-3809 - + 0 4869) differences between the correlation coefficients com- puted for intake of food in 1962-63 and 1947-48. DISCUSSION The correlation coefficient between gastric cancer In order to test our hypothesis we have chosen two mortality rates and the intake of animal proteins is countries with extreme gastric cancer mortality significant only for the intake during 1947-48, rates. The USA represents a country with one of the whereas, according to the test, it was doubtfully lowest, Czechoslovakia a country with one of the significant for animal protein in 1962-63. These highest gastric cancer mortality rates. Gut: first published as 10.1136/gut.10.12.1031 on 1 December 1969. Downloaded from

Gastrointestinal cancer and nutrition 1033 m 14 MORTALITY DATA These were available for both 0 sex CD 0*. sexes. As we found a very low variability in the 0 mortality from gastric and intestinal oC 10 0 ratio in the cancer, and as the correlation coefficients between the 0 male and female cancer mortality rates are strikingly ;- 6 0 cancer 4; high, 0.984 for gastric and 0.952 for intestinal, %.,- and Kurihara, 1966), we analysed the male 0 (Segi h2 0 mortality rates only. a 1947/48 A- 0 FOOD CONSUMPTION The analysis of food consump- 14 0 * E 0 tion between the USA and Czechoslovakia showed * 0 U, 0 was ulO * that the caloric intake in both countries practi- 0 cally the same. It was the structure of the food 0 0 consumed that interested us most. Here the com- U) parison showed a striking difference in the intake c 0 :_ 0 0 0 of vitamins A, B2, and C. Though it would have been 'D2 highly interesting to follow up these indicators, there ce . . . . .-a . . no available data for a com- 0 20 40 60 8( were unfortunately 1962/63 parison on an international scale. Animal protein intake (1.) Although the intake of fat is much higher in the USA than in Czechoslovakia, we could not draw FIG. 2a and b. Results of correlation analysis of intestinal any valid conclusions from the analysis of the fat cancer mortality rates related to normal protein intake. intake in the two periods studied on an inter- national scale. The data on the fat intake express only the direct intake, that is, not fat contained in other nutrients. Furthermore, the fat intake correlates highly with the intake of calories. Nor are the data for the intake of carbohydrates complete, as we have data on the intake of cereals only. It is noteworthy that the differences in the total http://gut.bmj.com/ protein intake between the USA and Czechoslovakia 70- were but slight. We therefore split up the total

W-4 on September 23, 2021 by guest. Protected copyright. Se' 20 S * * Another point of view also led us to consider the intake of animal protein as the basic indicator of - the qualitative structure of the food intake, namely, 1947/48 the high correlation between the intake of animal , 70 protein per capita and the expenditure for private 0 one ofthe main components E 60 consumption per head, of the gross national product (Dewhurst et al, 1961). L.. : 00 0 All these considerations led us to the correlations 0 0 .000 between the gastric and intestinal cancer mortality rates and the intake of animal protein. Initially, we 0 0 . 20 the 00 also worked out multiple correlations expressing u) the 0 relation between the number of calories, quantitative indicator, and the proportion of animal 0 20 40 60 8C indicator of food intake, to 1962 /63 protein, the qualitative Animal protein intake (/*I) the gastric and intestinal cancer mortality rates. Those multiple coefficients differed only slightly from FIG. 3a and b. Correlation between gastric mortality the simplecoefficients, asthere existsahighcorrelation rate and animal protein intake. between the intake of calories and animal protein. Gut: first published as 10.1136/gut.10.12.1031 on 1 December 1969. Downloaded from

1034 0. Gregor, R. Toman, and F. Prusova Therefore, we confined ourselves to the of Backett, E. M. (1967). The Epidemiology of Gastric Cancer. Report, study the Working Group, World Health Organization, Copenhagen. intake of animal protein only. Chaklin, A. V. (1962). Geographical Differences in the Distribution It should be said that our findings are based on the of Malignant Tumours, Bull. Wld. Hlth Org., 27, 337-358. Clemmesen, J. (1965). Statistical studies in the aetiology of malignant statistical data for mortality rates, and of course . Acta Path. microbiol. scand., suppl. 174, pts. we cannot exclude differences in diagnostic levels 1-2. in the Czechoslovakia (1966). Statistickd roeenka CSSR (Statistical Abstract different countries. As there are no other com- of Czechoslovakia) UKLKS, Prague. parable data either on incidence or on mortality Dewhurst, J. F., Coppock, J. 0., Yates, P. L., and Associates. (1961). available, we used those for our Europe; Needs and Resources. The Twentieth Century Fund, analysis. New York. As far as the problem of the time lag between the Doll, R. (1956). Environmental Factors in Aetiology of Cancer of the intake of nutrients and the mortality from gastro- Stomach. Gastroenterologia (Basel), 86, 320-328. Dunn, J. E., and Buell P. (1966). Gastrointestinal cancer among the intestinal cancer is concerned, we correlated the ethnic groups in California, III World Congress of Gastro- mortality rates for 1962-63 with the intake of animal enterology, Tokyo 1966. protein at two Fisher, R. A., and Yates, F. (1949): Statistical tables for Biological, different periods, because we Agricultural and Medical Research. Oliver and Boyd, London, supposed a time lag existed between the action of 1949. pathogenetic factors and the clinical manifestation Food and Agriculture Organization (1948-65). La situation mondiale of de l'alimentation et de l'agriculture, 1948, 1963, 1965. FAO. gastrointestinal cancer. Gregor, 0. (1965). Early Diagnosis of Gastric Cancer (in Czech). The correlation coefficients obtained for intestinal Czechoslovak Medical Press, Prague. , Toman, R., Pastorova, J., and Drnkovi. R. (1966). cancer are highly significant for both periods studied. Epidemiology of cancer of the stomach and other parts of the The correlation coefficient for gastric cancer is alimentary tract. III World Congress of , Tokyo 1966. significant only for the animal protein intake in Prusova, F., Drnkova, V., and Pastorova, J. (1969). 1947-48. This period has been chosen as it was the Geographical distribution ofstomach cancer in Czechoslovakia. first after World War II for which data on Gut, 10, 150-154. food Haenszel, W. (1958). Variations in Incidence of and Mortality from intake were available. with Particular Reference to the United The results of our statistical analyses do not con- States. J. nat. Cancer Inst., 21, 213-263. Higginson, J. (1966). Etiological factors in gastrointestinal cancer in tradict the presumption of a time lag as far as gastric man. Ibid, 37, 527-545. cancer is concerned and do not support its presump- Hirayama, T. (1963). A Study of epidemiology ofstomach cancer, with special reference to the effect of the diet factor. Bull. Inst. tion in the case of intestinal cancer. Publ. Hlth, 12, 85-96. Dunn and Bull (1966) also report similar Jones, F. A. (1966). The epidemiology of gastric cancer with special differences, but they arrive at reference to Causation. III World Congress ofGastroenterology, their conclusions from Tokyo 1966. an entirely different angle. They studied the change in Mider, G. B. (1960). Causative Factors in Gastrointestinal Cancer. gastric and intestinal cancer mortality rates in Gastroenterology, 39, 271-278. http://gut.bmj.com/ Sax6n, E. A., and Hakama, M. (1967). Cereal consumption and Japanese immigrants who had moved from a gastric cancer. Indian J. Cancer, 2, 265-268. country with a low standard of living to a country Schwartz, D., Lassere, 0., Flament, R., and Lellough, J. (1966). with a Alcohol et cancer. Europ. J. Cancer, 2, 367-372. high standard. They found a declining rate for Segi, M., and Kurihara, M.(1966). Cancer Mortality for Selected Sites gastric cancer and an increasing rate for intestinal in 24 Countries, (1962-1963). Dept. of Public Health, Tohoku cancer in these Japanese immigrants, University, Sendai, Japan. especially United States (1964). The Statistical Abstract of the USA. Dept. of marked in their offspring. Commerce, Washington.

Wynder, E. L., Kmet, J., Dungal, N., and Segi, M. (1963). An Epidemi- on September 23, 2021 by guest. Protected copyright. logical Investigation of Gastric Cancer. Cancer (Philad.), 16, 1461-1496. , Graham, S., and Eisenberg, H. (1966). Conference on the REFERENCES Etiology of Cancer of the . Report of the Research Committee W.H.O., on gastroenterology. Ibid, Acheson, E. D., and Doll, R. (1964). Dietary factors in carcinoma of 19, 1561-1566. the stomach, a study of 100 cases and 200 controls. Gut, 5, , Shigematsu, T. (1967). Environmental factors of cancer of the 126-131. colon and rectum. Ibid, 20, 1520-1561.