International Journal of Cardiology, 25 (1989) 333-338 Elsevier

333

CARD10 00980

Similarities and differences between the epidemiology and possible dietary causes of coronary arterial disease and strokes

Stephen Seely

Department of Cardiology, University of Manchester, The Rqval InfirmaT, Manchester, U.K.

(Received 11 May 1989; Revision accepted 27 June 1989)

Seely S. Similarities and differences between the epidemiology and possible dietary causes of coronary arterial disease and strokes. Int J Cardiol 1989;25:333-338.

The findings of a previous statistical study of the relationship between food consumption and mortality from coronary arterial disease are compared with those of a new study of food consumption and cerebrovascular disease.

The previous study found strong positive correlation between male mortality from coronary arterial disease and the consumption of two food items, milk and oats. Stroke mortality appears to be linked with the consumption of a wide variety of proteinaceous plants. These can be divided into two groups, one with an apparently strong and the other with an apparently mild effect. The strong group consists of the protein content of pulses, oats, maize and rye, the milder group of the protein content of wheat, rice, barley, potatoes and vegetables. The relative effect of the two groups of proteins is of the order of 6/l, but the proteins with a relatively mild effect are consumed in large quantities in comparison with those having a strong effect so that their effect is not negligible.

The correlation coefficient between male cerebrovascular mortality in the younger age groups and the combined consumption of the two groups of plant proteins is 0.91. That between male mortality from coronary arterial disease and the consumption of milk and oats found in the previous study was 0.94.

Key words: Atherosclerosis; Ischemic heart disease; Cerebral arterial disease.

Introduction

Coronary artery disease and cerebrovascular disease are thought to be essentially similar atherosclerotic disorders, affecting the organs most vulnerable to such disorders, the heart and the brain. Arterial disease, however, is of great com- plexity and non-uniform aetiology. In some in-

Correspondence to: S. Seely, Dept. of Cardiology, Univer-

sity of Manchester. The Royal Infirmary, Manchester Ml3 9WL, U.K.

stances, the greatest hazard presented by an atherosclerotic plaque is that it encroaches on the lumen of the artery and may ultimately occlude it. In another instance, the affected artery tends to bleed and ulcerate, possibly resulting in thrombo- sis. In yet another instance, it causes mechanical weakening of the arterial wall which may then rupture and haemorrhage under high blood pres- sure. Mechanical differences between the arteries of the heart and the brain may determine which of these hazards is most important under a given set of conditions. Thus. cerebral arteries are mechani- cally weak. In their case, high blood pressure

0167-5273/89/$03.50 0 1989 Elsevier Science Publishers B.V. (Biomedical Division)

334

together with any influence that may further weaken their walls is the most important hazard. The heart compresses its own arteries during sys- tole and has to rely on the maintenance of an adequate diastolic pressure for its blood supply. This makes it vulnerable to low blood pressure and arterial obstruction.

An example to illustrate the differences is that high serum cholesterol is a well-known risk factor of coronary arterial disease, while in cerebral arteries, as pointed out in a recent paper [l], low serum cholesterol level constitutes a risk. Very low serum cholesterol levels weaken the endothelium of cerebral arteries and. combined with high blood pressure, enhance the risk of haemorrhagic stroke. From such considerations, it could be expected that the causes of strokes and heart attacks would be similar in some respects yet different in others.

When the epidemiology of the two diseases is studied, the differences appear overwhelming. The geographical distribution of the two are so differ- ent that they might be entirely independent dis- eases. The prevalence of coronary arterial disease is highest in Northwest Europe and in English- speaking countries, that of cerebrovascular disease in odd, ethnically unrelated countries, like Bulga- ria. Portugal, Japan. Both diseases are compara- tively rare in third-world countries, but cerebro- vascular disease causes unexpectedly high mortal- ity rates in some tropical islands, like Mauritius, Martinique, Trinidad and Tobago, while coronary mortality in young males is high in India and Pakistan. Coronary mortality is highest in ad- vanced, industrialised countries, stroke mortality in some moderately prosperous countries, tending to decline with a further increase in prosperity. The best example is Japan, the world-leader in stroke mortality in the early 1960s where, coincid- ing with a phenomenal increase in prosperity, mortality from that cause halved in the last 25 years.

It must be noted, however, that a fortuitous element is often superimposed on such general patterns. sometimes giving rise to considerable differences between approximately equally pros- perous, possibly ethnically related neighbour countries. Thus. stroke mortality in Portugal, in round figures. is twice as high as in Spain and

three times as high as in France. In Scotland. mortality from both coronary and cerebrovascular disease is about a third higher than in England.

Nothing changes as rapidly and capriciously in prosperous countries as dietary habits, so that, if a disease causes higher mortality in advanced than in undeveloped countries, has a capricious geo- graphical distribution and sometimes gives rise to considerable swings in mortality in a compara- tively short time, connexion with food consump- tion may reasonably be suspected. The epidemio- logical peculiarities of stroke mortality, for in- stance, seem to point to a pathogenic agent associ- ated with an inessential food item, popular in some countries, less popular in others, but gener- ally tending to lose favour in the most prosperous countries. When attempting to evaluate the effect of dietary changes, however, we must not lose sight of the multifactorial aetiology of arterial diseases. Thus, a reduction of stroke mortality in a country could be the result of dietary changes, better medication (particularly the increasing use of anti-hypertensive drugs), change in smoking habits, or, most probably, of a combination of all these and possibly other factors. In spite of such difficulties, if the resulting changes in mortality are significant, it is usually possible to evaluate the effect of dietary changes by statistical methods.

If both coronary arterial disease and cerebro- vascular disease are connected with food. they are clearly not connected with the sume food, since, in that case, the large epidemiological differences between them would be unexplicable. This point is apparently not appreciated. because in current medical practice patients with coronary or cerebral arterial diseases receive the same dietary advice, namely to reduce their consumption of animal fats, and to increase that of fruit, fibres and the like.

The object of the present paper is to investigate the difference between the two diseases in respect of their connexion with food.

Methods

In a previous paper [2]. I suggested a statistical method to serve as a first approach to the possible causes of suspected food-related diseases. In a

335

group of countries, for which both mortality and food consumption statistics of an acceptable standard are available - for example, the 21 mem- ber countries of the Organisation for Economic Cooperation and Development - food consump- tion in the three or four countries with the highest mortality from a given disease is compared, item by item, with that in a similar number of countries with the lowest mortality. It can be assumed that

those food items, the consumption of which in the low mortality group e.uceeds that in the high mortality group, are unlikely to have a significant effect on the pathogenesis of the disease. They can then be eliminated from the list of suspects. Such a process of elimination finally produces a short list of suspect foods, the consumption of which is consistently high in the high-mortality, and low in the low-mortality group of countries. The items on this list then can be further investigated by other statistical methods, and can be offered for other branches of research for study. It is not claimed that this method finds all food items connected with a given disease, or eliminates all items uncon- nected with it, but it presents the results of sys- tematic search. not simply a guess, which has a reasonable change of finding the dietary factors of major importance relevant to the disease under investigation.

The suggested method can be applied to any suspected diet-related disease, disclosing negative as well as positive correlations. Subgroups within countries for which mortality statistics are availa- ble, such as males and females, or younger and older age groups, can be studied as well as whole populations.

The present paper examines only positive corre- lations. The source of mortality data are World Health Organisation Statistics Annuals [3], and those of food consumption data are the Food Consumption Statistics of the Organisation for Economic Cooperation and Development [4].

Results

I presented a statistical study on the described lines, investigating the connexion between male mortality from coronary arterial disease in 1984 and food consumption in 1976 in an earlier issue

of this Jourrzal [2]. Food consumption in the four countries of the Organisation for Economic Coop- eration and Development with the highest mottal- ity (Finland. Ireland, the United Kingdom and Sweden), was compared with that in four coun- tries with the lowest mortality (Spain, Portugal. France and Japan). Even such a simple study was sufficient to produce some surprising results. The consumption of eggs, for instance, the original suspect of the cholesterol theory, was 50% higher in Japan, the country with the lowest mortality. than in Finland, with mortality 8 times as high as in Japan. The correlation coefficient between male mortality from coronary arterial disease in the 8 countries above and the consumption of animal fats and proteins was 0.77 and 0.74. respectively, indicating a reasonable degree of positive correla- tion, but overshadowed by a correlation coeffi- cient of 0.9 between mortality and the consump-

tion of milk, and, even more surprisingly. of oats. More detailed investigation. involving 21 member countries of the Organisation for Economic Co- operation and Development confirmed the leading position of milk and oats. Trivariate analysis, with milk and oats as independent variables. gave a correlation coefficient of 0.94 and showed that. weight for weight, the apparent pathogenic effect of oats was 4.5 times as strong as that of milk proteins. The regression line of mortality from coronary arterial disease in older men against the combined consumption of milk and oats is shown in Fig. 1.

A similar study, attempting to discover connex- ions between cerebrovascular disease and food consumption, immediately demonstrated greater difficulties than those in the investigation of coronary arterial disease. No food item can be found which stands out with strong correlation with mortality.

Preliminary investigation has shown that the ranking order of the member countries of the Organisation for Economic Cooperation and De- velopment in respect of mortality from strokes is different in the younger and older age groups. Thus, the current world leader in stroke mortality in the 35544 age group is Hungary. in the 45554 age group Mauritius, while in the older age groups Bulgaria recently superseded Portugal. When

336

!Y T . Japan

f 0 y=-0.44x2 + 53.2x + 153

I I I I 0 10 20 30 40 50

COMBINED CONSUMPTION OF UNFERMENTED

MILK PROTEINS AND OATS

Fig. 1. Male mortality from coronary arterial disease in the 65-74 age group in 21 OECD countries, in relation to the

combined consumption of milk proteins and oats. See text.

Reproduced from [2].

mortality rates change, they may not change equally in the younger and older age groups. For instance, the recent decline in mortality in Japan was mainly in the older age groups, leaving mortality in the younger groups still very high. This phenomenon can be interpreted by assuming that strokes are associated at least with two food items (or two groups of food items), one with a quicker action the other with a slower, cumulative effect. If the assumption is correct, it might be easier to identify the quicker-acting agent. It was decided, therefore. to make a start with the study of a younger (45-54) age group, presumably more under the influence of this faster agent than those dying of strokes in old age.

As before, mortality data for the year 1984 were correlated with food consumption in 1976. In this study, the high mortality group for cerebro- vascular disease consisted of three countries (Japan, Portugal and Yugoslavia) and the low mortality group of Switzerland, Canada and Sweden.

Comparison of the two groups immediately re- veals the fact that the consumption of foods of animal origin (meat, animal fats, poultry, eggs,

milk and other dairy products) is substantially higher in the low-mortalit_v group. Thus, the aver- age consumption of animal fats in Switzerland, Canada and Sweden (109 g per person per day) is nearly twice as high as that in Japan, Portugal and Yugoslavia (58 g/day). The same applies to animal proteins, 70 and 36 g/day, respectively. The only exception is fish, with an erratic consumption both in the high-mortality and low-mortality group (103.1, 89.1 and 8.5 g/day in Japan, Portugal and Yugoslavia, and 14, 20.1 and 62.9 g/day in Switzerland. Canada and Sweden). Generally, the correlation between stroke mortality and the con- sumption of most animal foods is strongly nega- tive and between it and fish consumption near zero. Among plant foods, the same applies to sugar, the consumption of which is substantially higher in the low-mortality group.

Among other foods of vegetable origin. the correlation is best with plant proteins. Thus, the consumption of all plant proteins in 1976 in the high mortality group (Japan, Portugal and Yugos- lavia) was 56.4, 46.2 and 65.2 g/day. In the low-mortality group it was 32.7, 34.2 and 26.6 g/day. The protein content of pulses (including soya beans) was also consistently higher in the high mortality group (12.4, 3.1 and 4.6 g/day) against 0.5, 0.4 and 0.5 g/day in the low mortality group. Lastly, an item with a higher consumption in the high mortality group was the protein con- tent of cereals (including rice), the two sets of figures being 26.8, 31.7 and 53 in the high mortal- ity group, 22.1, 21.4 and 18.4 in the low mortality group. When the consumption of these three items, (the protein content of all plants, pulses and cereals) was examined in the 21 countries of the Organisation for Economic Cooperation and De- velopment, correlations were unimpressive. Thus, the correlation coefficient between stroke mortal- ity and the total consumption of plant proteins was 0.68, between it and the protein content of pulses 0.67. and between it and the protein con- tent of cereals 0.60.

The results seem to point to some agent widely distributed in proteinaceous plants and, hence, to great difficulties in identifying it by statistical analysis. The correlation between stroke mortality and the consumption of an individual food item,

337

P

which supplies only a small part of the total intake of a hypothetical pathogen, is likely to be weak, probably indistinguishable from positive correla- tions arising by chance. Correlation with the total intake of plant proteins is weak, because the major part of the total intake is harmless.

The following technique was found to provide a way out of this difficulty. The food consumption statistics used in this study [4] list 12 plant foods containing significant amounts of proteins: wheat, rye, oats, barley, maize, rice, pulses, potatoes, nuts, vegetables, fruit and cocoa. Each of these items was, in turn, excluded from the total intake of plant proteins, to check whether its omission made the correlation coefficient between stroke mortal- ity and the remaining total of plant proteins better or worse. Those items, the omission of which improved correlation, were deemed harmless and excluded from the suspect list; those which made correlation worse constituted the final list of sus- pects. In the course of this work it was found that the omission of some items made a considerably bigger difference per unit weight than of others, presumably indicating a stronger pathogenic ef- fect. The final result was that three items, (the protein content of fruits, nuts and cocoa) were excluded from the suspect list, and the remaining 9 could be divided into two groups. The first group, with an apparently strong effect, consisted of pulses, rye, oats and maize. The second group, with an apparently milder effect, comprised wheat, barley, rice, potatoes and vegetables. The dif- ference between the apparent pathogenicity of the two groups was of the order of 6/l.

This work was rewarded by a large increase of the correlation coefficient. Trivariate analysis be- tween male cerebrovascular mortality in the 45-54 age group in 1984, and the consumption of the two groups of plant proteins yielded the equation

Y = 3.55~~ + 0.58~~ - 1.16,

where Y is mortality rate per 100,000 living popu- lation, and xi and x7 are the per capita consump- tion, in grams per day, of the two groups of plant proteins, respectively. The correlation coefficient is 0.91. The relative potency of the two groups of plant proteins is indicated by the ratio of the

: 0-J / I I I I I I

0 20 40 60 80 100 120

COMBINED CONSUMPTION OF PLANT PROTEINS

Fig. 2. Male stroke mortality in the 45-54 age group in 21

OECD countries. in relation to the combined consumption of

plant proteins. See text.

constants 3.55 and 0.58, which gives its value as 6.1. Fig. 2 shows the regression line in which mortality is plotted against the combined con- sumption of the two groups of proteins. the pro- teins with stronger effect being weighted by a factor of 6.1 and added to the proteins with a milder effect.

In this work, a time interval of 8 years was allowed between food consumption and mortality data. Other time intervals between 4 and 12 years were tried but found to make little difference.

The correlation with the consumption of the two groups of plant proteins is also fairly strong in older men, the correlation coefficient being of the order of 0.8. The relative potency of the two groups of plant proteins appears to change in the older age groups, the previously found factor of 6.1 decreasing to 5.2 in the 55-64 age group and to 4.7 in the 65-74 age group. This presumably indicates variations in individual sensitivity to some pathogenic agent associated with plant pro- teins having a strong effect, people of high sensi- tivity tending to die at an early age if their con- sumption of the critical food items is high.

Discussion

It is interesting to note that, in spite of the great differences between the epidemiology of

33x

coronary arterial disease and cerebrovascular dis- ease. there is at least one plant food, oats, which is apparently connected with both.

I can hazard no guess as to what kind of pathogenic agents relevant to arterial disease might be present in proteinaceous plants, apart from the general observation that all plants defend them- selves by the production of secondary plant metabolites which are toxic to animals. The study of plant toxins, identifying literally hundreds, if not thousands, of secondary plant metabolites, some of considerable subtlety and potency, has a large literature [5-71. Many plant toxins are targeted on the circulatory system of animals, as blood coagulants and anticoagulants, vasodilators and vasoconstrictors. It is hoped that the present paper will initiate further study of the possible toxins in pulses and cereals, particularly in oats. rye and maize.

No mention was made in this study of salt consumption. an excess of which tends to increase

blood pressure and thus probably constitutes a risk factor for strokes. The reason is that the food consumption statistics of the Organisation for Economic Cooperation and Development do not list salt as a food item. In any case, the wastage of table salt in prosperous countries is such that statistics are likely to be unreliable, and statistical analysis could add nothing useful to existing liter-

ature. It has already been pointed out that, in present

clinical practice, coronary and stroke patients re- ceive the same dietary advice, notably to reduce their intake of animal foods and increasing that of

plant foods. Such advice may be of some value to patients with coronary arterial disease, but stroke patients are directed by it exactly to those pro- teinaceous plants which they may do well to avoid. If the findings of this paper are should reduce their intake of plant sent in pulses, oats, rye and maize.

correct, they proteins pre-

Acknowledgements

My thanks are due to the editor of the Journal, Professor Robert H. Anderson. for encourage- ment, and to the referee of the paper, Dr. D. Hywel Davies. for constructive criticism and welcome suggestions for improvements.

References

Iso H. Jacobs DR Jr. Wentworth D. Neaton JD, Cohen JD.

Serum cholesterol levels and six-year mortality from stroke

in 350,977 men screened for the Multiple Risk Factor Inter-

vention Trial. N Engl J Med 1989;320:904&910.

Seely S. Diet and coronary arterial disease: a statistical

study. Int J Cardiol 1988:20:183%192.

World Health Organisation. World Health Statistics An-

nuals. Geneva. 1983-1986.

Organisation for Economic Cooperation and Development.

Food Consumption Statistics, 1964-1978. Paris 1981.

Liener IE. ed. Toxic Constituents of Plant Foodstuffs. New

York: Academic Press. 1980.

Roaenthal GA. Janren DH. eds. Herbivores: their interac-

tion with secondary plant metabolitea. New York: Academic

Press. 1979.

Silverstone GA. Possible sources of food toxicants. In: Seely

S. Freed DLJ. Silverstone GA. Rippere V. eds. Diet-related

diseases: the modern epidemic. London: Croom Helm. 1985.