Chapter Vlll

GENERAL DISCUSSION AND CONCLUSIONS

BODY FAT AND CORONARY ATHEROSCLEROSIS

As mentioned in the introduction, atherosclerosis is believed to be due to a number of different factors. It was also pointed out that the investigation included analysis of the correlation between atherosclerosis and one of the various factors discussed, namely body-build. Since a variety of clinical investigation and autopsy studies suggest the existence of a relationship be- tween the amount of body-fat and atherosclerosis (WILENS 1947, DUBLIN et al. 1952, HENSCHEN 1959, SANDERS 1959 and others), attempts were first made to determine whether or not such a correlation could be demonstrated. A statistically significant correlation ( r= +0.32**) was found.

But the demonstration of such a correlation does not forthwith imply any causal relationship. Decision of that question requires further analysis with biologic evaluation of the various alternatives that might possibly be respon- sible for such a correlation.

A correlation between variables CI and b might thus be due to variations in b being caused directly by variations in CI. Then, one may speak of a direct causal relationship. One might also imagine that the correlation might be due to CI influencing a third variable c, which in turn influences b. Then one may speak of a mediated causal relationship. Finally, the correlation may be of indirect nature, c( and b being influenced by a third factor c in the same or opposite directions.

A correlation of possibly direct causal nature between the amount of body- fat and atherosclerosis might be due to an increased amount of body-fat increasing the strain on the vessel walls and thereby accelerating athero- sclerosis. A causal relationship might also theoretically be mediated by the blood pressure, i. e. an increased amount of body-fat might result in hyper- tension, which in turn might cause atherosclerosis (see KEYS 1954). A n in- direct correlation might be of two types, dependent on genetic or environ- mental factors. If it were due to hereditary factors, it would imply a genetic link between disposition to obesity and disposition to atherosclerosis. If, on the other hand, the factor influencing both the amount of body-fat and atherosclerosis were of environmental nature, it would imply that the amount of body-fat and the severity of atherosclerosis vary with nutrition and related

86

factors. Finally, one might imagine a combination of two or more of the alternatives discussed to be responsible for the correlation.

The interplay reflected by the statistically demonstrated correlation is of importance in preventive medicine. If a correlation between the amount of body-fat and atherosclerosis were of direct causal nature, one might imagine that reduction of the amount of body-fat would retard the progress of that part of the atherosclerosis due to such a correlation. On the other hand, if the correlation were due to nutrition, preventive measures could be taken by correcting the diet. If, however, the correlation were due to hereditary factors, it is questionable, whether preventive measures could be taken.

Analysis of the supposed relationship between state of nutrition and athero- sclerosis has been investigated from various angles. For example, the decrease in atherosclerosis and manifestations of atherosclerosis during times of war have been studied by various authors (VARTIAINEN et al. 1947, MALMROS 1950, ZSCHOCH 1957 and others). It is, however, difficult to decide with cer- tainty what factor or factors is responsible for any decrease during such abnormal conditions as are prevalent in times of war. Comparative ende- miologic investigations in different parts of the world where food is abundant and where food is scarce have shown atherosclerosis to be common where food is abundant, and less common where it is scarce (SNAPPER 1941, WILENS 1947, TEJADA et al. 1957). These findings have been discussed, and it has been shown that several factors other than dietary might have been of significance (MANN 1957). Studies of this type will thus not permit proper evaluation of any such correlation. It might, however, be possible by a long- term investigation of morbidity and mortality from atherosclerosis in two groups with an initially equal amount of body-fat subsequently reduced in one group by dietary measures (compare the fall in mortality of insurance policy holders in America, who reduced their relative body-weight (DUBLIN & MARKS 1952)). The effect of environmental and genetic factors might also be studied in related material, such as classical twin investigations. It would, however, require a long time to perform such an evaluation according to these principles and it would also be difficult to collect such a material.

In the present investigation the problem was approached from another angle: a cross-sectional study of autopsy material.

For a correlation between the amount of body-fat and severity of athero- sclerosis to be analysed in such a material, however, it is necessary to distin- guish between components, mainly dependent on environmental influence (labile) or mainly on genetic influence (stable), in the fatty tissue.

It is known that the amount of body-fat varies with environmental condi- tions. Certain experimental investigations (HAUSBERGER 1955) and obser-

vations on genetically obese, yellow mice (DANFORTH 1927, DICKERSON et al. 1947), however, argue for the existence of genetic factors, too. Observations on fatty tissue transplants in human beings (STRANDBERG 1915, HOFF 1953), and correlation demonstrated between hair growth on the body and depth of the panniculus adiposus, as judged by skin fold calipers (LINDEGARD et al. 1956), directly support such an assumption.

In the present investigation the weighted mean value of the corrected thick- ness of the subcutaneous fat in three sites of the panniculus adiposus was used as an expression of the total amount of body fat. The choice of these sites and the interpretation of the findings were made in accordance with the recommendations of BROZEK & KEYS (1951), KEYS and BROZEK (1953) and PASCALE et al. (1956).

The amount of body-fat, as estimated by the fat thickness, showed a corre- lation with hair growth. This correlation is apparently not of direct causal nature but due to a third factor influencing both variables. Since hair growth on the 'body is dependent mainly on genetic factors (REYNOLDS 1951), this third factor might be dependent on the gene pattern of the individual. As already assumed, the correlation between fat thickness and hair growth thus implies that the inter-individual variation in body fat may be explained in part by variation in a stable fat factor. The fat thickness also varied with a mainly labile part of the muscle factor and with age. This implies the exi- stence of a labile component. The results obtained in the present investigation thus support the assumption that the variation in the amount of body-fat is due to variation in a stable and a labile factor.

Morphologically, the variation in the amount of body-fat may be ascribed to variation of two factors, namely size and number of fat cells. Earlier inves- tigations on geese (CLARA 1929) and on man (HAAGENSEN 1953, ENTENMAN et al. 1957) have shown that the size of the fat cells is dependent on the state of nutrition of the individual. Theoretically, it might then be assumed that the size of the fat cells is mainly an expression of the labile fat factor, while the fat celI number is mainly a representative of the stable fat factor of the individual. The validity of this hypothesis is tested below.

The size and number of the fat cells in specimens of fat removed post mortem from those sites where the fat thickness was measured was deter- mined according to a specially devised method. Statistical analysis showed that the cell count varied with cell size. To study the variation in the cell count independently of cell size, the influence of the cell size on cell count must be eliminated. For purely technical reasons, the fat thickness was used instead of the cell count, a procedure which was considered justified because the correlation between cell count and fat thickness was highly significant

88

(r=+0.98**), and the two variables may thus be regarded as varying in an identical manner. This elimination was secured by the residual method (page 51). As a final expression of the cell number the residual F fat thickness (cell size) was used. Statistical analysis showed that the cell size was responsible for the correlation between fat thickness and the mainly labile muscle factor and with age, while cell number, expressed as the residual F fat thickness (cell size) was responsible for the correlation between fat thickness and body hair growth. The results thus directly support the hypothesis that the cell size really represents u measure of the labile component of fat thickness, while the cell number represents u measure of the stable component of fat thickness.

The correlations between the labile and stable fat factors on one hand and coronary atherosclerosis on the other showed that the co-variation of the fat thickness with coronary atherosclerosis (r= +0.31**) could be ascribed en- tirely to the variation in the labile fat factor, i . e . cell size (r= +0.32**). The stable fat factor, i. e. cell number was not found to vary with the severity of coronary atherosclerosis.

The results of the investigation thus permitted a biologic evaluation of the correlation between fat thickness and coronary atherosclerosis. It does not appear likely that the correlation was of direct causal nature. If it had been, the correlation between the fat thickness and coronary atherosclerosis would presumably have been due to both components of the fat thickness (compare correlation between fat factors and heart weight, page 82). In order to deter- mine whether any causal relationship might be mediated by the blood pressure (as judged from heart weight in the present investigation), a special analysis was performed, which, however, revealed nothing suggesting the existence of any such mediated causal relationship. The correlation could, however, be referred to the variation of only one component of the fat thick- ness, namely the labile factor. This suggests that the correlation was of indi- rect nature and due to nutrition and related factors.

Atherosclerosis o f the coronary branches was thus found to be related with the morphology of the fatty tissue. The findings support the hypothesis that it is mainly the environmental factcrs in the accumulation of fat thrit (ire responsible for the correlation between the amount of body-fat and coronary atherosclerosis.

The results thus suggested the existence of a correlation between amount of body-fat and coronary atherosclerosis being due to a factor accessible to preventive medicine.

89

BODY BUILD WITH THE EXCEPTION OFBODY FAT AND CORONARY ATHEROSCLEROSIS

Since opinions also differ on the correlation between atherosclerosis and body-build in general, other components of body-build were studied for any correlation with coronary atherosclerosis.

The divergence of opinion is at least to some extent due to the methods used for estimating atherosclerosis and body-build permitting different inter- pretations of the results. It might therefore be convenient first to discuss the validity of different methods for assessing atherosclerosis and body-build.

The severity of atherosclerosis of the vessels in internal organs can as yet only be judged with certainty at autopsy. Then, direct examination of the vessels is possible. Attempts have been made to assess the degree of athero- sclerosis by chemical analysis with determination of the calcium, total lipoids and cholesterol content in the entire vessel or in excised parts of the vessels. BJ0RNssoN (1941) made a critical analysis of the chemical methods for assessment of atherosclerosis. He concluded that they did not permit proper estimation of the severity of atherosclerosis in the vessels. An essentially similar conclusion was arrived at by WHO Study Group on Classification of Atherosclerotic Lesions (1958). The Study Group concluded that atheroscle- rosis should be graded essentially according to its gross appearance. The basis for a grading system outlined by the Study Group had, however, already been applied by SJOVALL & WIHMAN (1934) in their well defined classification scheme. Their classification scheme thus appears to be suitable for evaluation of the severity of atherosclerosis.

The inter-individual variation in body-build is at present assessed mainly by means of two methods, the relative weight (weight in relation to stature), which, with certain modifications, is used by the America1 life insurance companies, and Sheldon’s somatotyping system.

The relative weight is, however, as mentioned, a complex term embracing variation in skeletal, muscle, and fatty tissue. A correlation in which one of the members is represented by such a complex term is, however, difficult to ana- lyse. The relative weight is therefore hardIy suitable for registration of variation in body-build. Even the variables in Sheldon’s system, which are estimated subjectively, are complex terms. These variables are assumed to define the permanent (constitutional) body-build of the individual (SHELDON 1940). This assumption has, however, been checked by LASKER (1947), who found that body-type determined according to Sheldon was not a permanent feature. Also the theoretical background of Sheldon’s typing system has been criticised (HUNT jr. 1949, KEYS et al. 1953). When body-build is classified

90

according to LINDEGARD (1953), the variation of sturdiness, length, muscle and fat factors, are determined. These components are objectively measured continuous variables of biologic significance. They can thus be analysed both statistically and biologically. This method was therefore used for analysis of body-build in the present investigation.

The skeletal factors are determined in accordance with LINDEGARD’S rating system. In the present investigation, however, the measurement used by LINDEGARD for assessing the muscle factor, the physiologic cross-section of the muscle, could not be employed, because the investigation was carried out on autopsy material. Therefore, the morphologic girth of the muscle was used instead. Since the strength of a muscle, as known, is proportional to its so-called physiologic cross-sectional area and since the increase in muscle mass in the adult is due mainly to increase in the thickness of the individual muscle fibres (HJORTSJO 1959), the strength of the muscle should also be proportional to the morphologic girth. The strength of the different muscle groups has also been found to be proportional to the total musculature, as estimated by creatinine excretion in the urine (LINDEGARD 1953). There- fore, the morphologic girth of a large muscle might be taken as a represen- tative of the muscle factor. Analysis showed that the muscle factor is corre- lated with the sturdiness factor but not with the length factor. This is in agreement with LINDEGARD’S earlier finding (1955). The muscle factor regu- larly showed a decrease with age in the age groups studied. This supports the tendency found by REYNOLDS et al. (1950) and NEWMAN (1952). Analysis of the correlation between the biceps girth and the skeletal factors and age thus supports the assumption that the biceps girth really is a measure of the muscle factor.

Analysis of the correlation between atherosclerosis and these factors showed that the grade of atherosclerosis in the aorta or coronary branches did not vary with the skeletal factors. Coronary or aortic atherosclerosis did not vary with the muscle factor either. On elimination of the effect of age, however, a significant partial correlation was found between the muscle factor and coronary atherosclerosis. This partial correlation proved to be partly dependent on the influence of the labile fat factor. On elimination of the latter, the partial correlation of second order, i. e. the correlation between the muscle factor and coronary atherosclerosis independent of the influence of age and the labile fat factor was only probably significant. On analogous analysis the muscle factor showed no partial correlation with aortic athero- sclerosis.

The correlation between body-build and coronary atherosclerosis was thus largely dependent on the variation of the labile fat factor, but to a minor extent also on variations in the musculature.

91

I

I \ \ \

I \

. . . . W I

I I

\ / \ I / / \ \

r,gnrhcaot corrcl.t,~"

- probobty s,gn#hcent corrdotmn - - no corr.tot,on . . . . no comlolion tort.foCtJ

Fig. 31. Correlation between coronary and aortic atherosclerosis and body-build factors.

92

CORONARY AND AORTIC ATHEROSCLEROSIS

Aortic atherosclerosis has often been taken as a measure of atherosclerosis in general or in the coronary vessels (ROSENTHAL 1934, FABER et al. 1949, ELEELES 1957 and others).

In order to check whether this is justified, a comparison was made in the present investigation between atherosclerosis in the different regions of coronary branches as well as in different regions of the aorta. It was found that the grade of atherosclerosis varied largely in parallel in the three coron- ary branches. It was also found that the grade of atherosclerosis in coronary branches was not significantly correlated with that in the ascending aorta (r= f0 .19) and in the thoracic aorta (r= f0 .20* ) . The grade of coronary atherosclerosis agreed better with that in the aortic arch (r=+0.35**) and in the abdominal aorta ( r= +0.44**), The correlation, however, was not so high that the grade of atherosclerosis in the coronary arteries could be pre- dicted from that in the aorta.

It thus appears that it is not justified to use the variation in atherosclerosis in the aorta as a measure of coronary atherosclerosis.

This assumption was also supported by the findings in the present investi- gation in which coronary atherosclerosis, but not aortic atherosclerosis, was found to vary with the amount of body-fat.

Thin might at least partly explain the contradictory results (according to BROZEK & KEYS 1957) obtained in autopsy studies regarding the correlation between obesity, usually as judged by relative body-weight, and atheroscle- rosis. The investigations referred to by BROZEK et al., in which no correlation was found, were based on examination of aortic atherosclerosis (ROSENTHAL 1934, FABER et al. 1949), while those who did find a correlation had based their study on atherosclerosis in general as assessed by the pathologist in his autopsy report or only on coronary atherosclerosis (WILENS 1947, LOBER 1953).