Comparison of Effects of Daily Physical Activity , Obesity , and Alcohol Drinking and Cigarette Smoking Habits on Serum Levels of High Density Lipoprotein Cholesterol and Apolipoprotein A-1 among Schoolchildren and Their Parents

Serum levels of major constituents of high density lipoprotein (HDL), HDL cholesterol (HDL-C) and apolipoprotein A-1 (Apo A-1), have been reported to be inversely related to occurrence of atherosclerotic heart disease1-4) and also reported to be affected by life style factors such as physical exercise 5-10) alcohol drinking5,9,11-16), cigarette smoking11,17-22), and obesity11,23). However, relative strength and qualitative differences of the effects of these factors have not been settled yet. Furthermore, there has been little report in which these characteristics were compared between adults and children. Elucidation of these characteristics would provide a clue to make efficient way of public education for preventing atherogenic lipid change. The purpose of this study is to quantitatively estimate relative strength of the effect induced by each of the above life style factors which ranged within the usual daily life in middle aged men and women and schoolchildren. Quality difference of HDL induced by these factors, which could be inferred when both HDL-C and Apo A-1 levels were available, was also discussed by referring respective changes of HDL-C and Apo A-1.

Serum levels of major constituents of high density lipoprotein (HDL), HDL cholesterol (HDL-C) and apolipoprotein A-1 (Apo A-1), have been reported to be inversely related to occurrence of atherosclerotic heart disease1-4) and also reported to be affected by life style factors such as physical exercise 5-10) alcohol drinking5,9,11-16), cigarette smoking11,17-22), and obesity11,23).However, relative strength and qualitative differences of the effects of these factors have not been settled yet.Furthermore, there has been little report in which these characteristics were compared between adults and children.
Elucidation of these characteristics would provide a clue to make efficient way of public education for preventing atherogenic lipid change.
The purpose of this study is to quantitatively estimate relative strength of the effect induced by each of the above life style factors which ranged within the usual daily life in middle aged men and women and schoolchildren.
Quality difference of HDL induced by these factors, which could be inferred when both HDL-C and Apo A-1 levels were available, was also discussed by referring respective changes of HDL-C and Apo A-1.Every subject filled out the form which asked the times when the subject attached and took off the device and the total energy daily expended by physical activity.The total energy measured for two days was averaged and divided by body weight, and this value was used in this study so that mean strength rather than total amount of physical activity could be related to changes of HDL constituents.

Materials
Drinking and smoking habits were inquired by a questionnaire having categorical answers and allocated scores as follows."How much do you drink Sake (a traditional Japanese alcoholic beverage brewed from rice) a day on the average?"The choices and scores of the answer were : none: 0, less than one unit : 0.5, one to less than two units: 1.5, two to less than three units : 2.5, and three or more units : 3.5, where one unit, a traditional measure of fluid volume equal to 180 ml, of Sake contained about 22 g of ethanol ."How many numbers of cigarette do you smoke a day on the average?"The choices and scores of the answer were : none : 0, ten or less : 5, 11 to 20 : 15, 21 to 40 : 30 , and 41 or more : 50.
To check consistency of Sake drinking with general drinking habit, the subjects were examined for the kind and amount of alcoholic beverages they took during two consecutive days in October by dietitians .The correlation coefficient between the scores of categorical answers and the amount of ethanol taken from any kind of alcoholic beverages during the survey was 0 .57(p<0.001) in 116 fathers out of 124.

STATISTICAL ANALYSIS
To estimate the effect of each of alcohol drinking , cigarette smoking, physical activity and BMI on HDL -C and Apo A-1 by adjusting confounding between these factors, first, multiple regression equation was obtained by using serum levels of HDL-C or Apo A -1 as the dependent variable and scores of categorical answers or measured values of the four life style factors as the explanatory variable .In the children and mothers, who had no smoking or drinking habits , only strength of physical activity and BMI were used as the explanatory variable.
Age of the subjects was not included as an explanatory variable since most parents were within the short age-range between late thirties and early forties.
Then, to make the effects due to the above four factors which have different measurement units comparable under the same condition, change of HDL-C and Apo A-1 to be produced when the factor under consideration was changed from the mean of the bottom tertile subgroup to that of the top tertile one was obtained as the difference of the expected values between the respective tertile subgroups which were calculated through the mutiple regression equation determined above.In this calculation, for the values of factors which were left out of consideration were entered the respective means of whole fathers, mothers or children.

RESULTS
The means and standard deviations of the variables of the subject groups are shown in Table 1.Strength of physical activity was highest in the children and lowest in the mothers.BMI was highest in the fathers and lowest in the children.
HDL-C was highest in the children and lowest in the fathers.Although Apo A-1 was highest in the fathers and lowest in the children, difference of Apo A-1 levels between the subject groups was smaller than that of HDL-C.
Correlation coefficient (R) of HDL-C and Apo A-1 in 93 triplets within the whole subjects is shown in Table 2. Correlation coefficient of HDL-C was significant between mothers and children (R=0.41,p< 0.01) but was insignificant both between fathers and children and between the parents.Apo A-1 took a positive, borderline significant correlation between children and either of the fathers and mothers (R= 0.19, p<0.10) but did not show a significant correlation between the parents.
When the multiple linear regression analysis was applied, significant relationship of the life style factors with HDL-C (coefficient of determination (R2)=0.29,p<0.01) and with Apo A-I (R2=0.23,P<0.01) was obtained for the fathers, and borderline relationship with HDL-C (R2 =0.03, p<0.1) and with Apo A-1 (R2=0.04,p<0.1) was obtained for the children as shown in Table 3.However, there was no significant or even borderline relationship for the mothers.When total amount of physical activity was used on behalf of strength of it, regression to HDL-C and Apo A-1 became slightly worse in the fathers, although there was no difference in significance level.In the children, however, borderline relationship disappeared.When only strength of physical activity and BMI

DISCUSSION
Total variation of HDL-C and Apo A-1 of the fathers were explicable as much as 29 and 23 percent, respectively, by strength of physical activity, obesity, and alcohol drinking and cigarette smoking habits in daily life.These values was higher than 18 percent reported by Fraser et al5).who investigated relationships to HDL-C in male adults by a follow-up study by adding fish consumption to the four factors similar to ours, and was also higher than 13 percent reported by   Apo A-1 shown in Figure 1.
Table 4. Differential of high density lipoprotein cholestrol (AHDL-C) and that of apolipoprotein A-1 (AApo A-1) and their 95% confidence intervals' expected by varying the level of each variable from the mean of the bottom tertile to that of the top one and the ratio of the differentials .necessarily provide a protective effect against atherogenesis to those in whom HDL3 subfraction was elevated.Therefore, epidemiological studies that take HDL subfractional concentration into consideration is necessary to determine whether alcohol is truly protective through lipid metabolism.
Unless an easy method to determine subfractional concentration is developed, HDL-C and Apo A-1 levels would be a useful proxy.
In the children, only several percent of total variation of HDL-C and Apo A-1 could be explained by strength of physical activity and obesity.Regression was even insignificant in the mothers.These results were compatible with the finding from the fathers that only 10 and 5 percent of total variation of HDL-C and Apo A-1, much smaller than that explained by the four factors, could be explained by combination of these two factors.HDL-C concentration of the children had a correlation with that of mothers with a coefficient of determination of 0.17, much higher than the coefficient in the regression by environmental factors, 0.03.This result may suggest that children of this age are affected more by a genetic factor than environmental factors.Although the correlation coefficient of HDL-C between mothers and children was higher than that of Apo A-1, the reason for this difference is beyond the scope of this paper and will be discussed in the following paper.
Bordeline relationship of BMI with HDL-C was obtained in the mothers in agreement with Haarbo et al.11) and others27) who showed that obesity was significantly related to decrease in HDL-C and Apo A-1 in post and premenopausal women.However, we could not obtain a significant relationship of physical activity to HDL-C and Apo A-1.One of the reasons for this negative finding would be due to a fact that the level of physical activity was lowest in the mothers and it might have been too low to affect them.Another reason would be shortness of subject number.Effect of nutrient intake such as carbohydrate on Apo A-1 has been reported35).So the factors such as dietary habit and sex hormons may have to be included in the analysis to find a better relationship in the mothers.Especially in children, effect of genetic heritability may have to be taken into consideration to obtain more significant relationship.

Figure 1 .
Figure 1.Expected serum high density lipoprotein cholesterol (HDL-C) and apolipoprotein A-1 (Apo A-1) concentrations and their ratio of the fathers when each of strength of physical activity, body mass index (BMI), and alchol drinking and smoking habits took the mean of the bottom (T1), the middle (T2) and the top (T3) tertile subgroups.Expected values were calculated through the multiple regression models shown in the Table 2.

Figure 2 .
Figure 2. Expected serum high density lipoprotein cholesterol (HDL-C) and apolipoprotein A-1 (Apo A-I) concentrations and their ratio of the 5th grade schoolchildren when each of strength of physical activity and body mass index (BMI) took the means of the bottom (T1), middle (T2) and top (T3) tertile subgroups.