Allele Frequency of Apolipoprotein Gene Polymorphisms and Association between Genotype and Serum Lipid and Apolipoprotein Levels

The rare allele frequency of the restriction fragment length polymorphism (RFLP) with Xba I (X+) at the apolipoprotein B (apo B) gene locus was 0.041 in men and 0.026 in women in H-Y district, Shiso County, Hyogo Prefecture, Japan : EcoR I at apo B gene (E-)=0.073 and 0.076, Msp I at apo Al-CIII gene (M-)=0.423 and 0.430, and Sac I at apo Al-CIII gene (S+)= 0.309 and 0.349. There was no marked ageor sex-difference in the frequencies. The frequencies of X+ and Ewere lower and those of S+ and Mwere higher in H-Y district than in Caucasian populations. In this population, according to the ANOVAs, the genotype for EcoR I was significantly (p<0.05) associated with serum total cholesterol, LDL cholesterol and apo B in women, and the genotype for Sac I with serum triglycerides and Msp I with serum apo CIII in men. The absolute values of Spearman correlation coefficients between genotypes and serum lipids or apolipoproteins were less than 0.2 after adjustment for age. J Epidemiol., 1995; 141-151.

Serum lipids and apolipoproteins are considered to be influenced by both environmental and genetic factors. In many epidemiological and experimental studies, it has been shown that several lifestyle factors are related to the level of serum lipids : e.g. a positive or inverse relation of total fat, saturated fat, cholesterol and polyunsaturated fat to serum total cholesterol (TC) and low density lipoprotein cholesterol (LDL-C)'), and a positive relation of physical activity2t and alcohol consumption') to serum high density lipoprotein cholesterol (HDL-C). Although, as for the genetic factors, there have been some family and twin studies as well as studies on the association of family history with hyperlipidemia, they have not only advantages but also disadvantages, e.g. "family aggregation" in family studies"), "shared environment" in family and twin studies'), and recall bias in family histories. In contrast to these studies, "candidate gene" approach by identifying genotypes related to inter-individual variations of serum lipids and apolipoproteins in populations has recently been proved highly effective with development of the polymerase chain reaction (PCR) method. One of the two aims of the present study is to epidemiologically describe the allele frequencies of four restriction fragment length polymorphisms (RFLPs) with Xba I and EcoR I at the apolipoprotein B (apo B), and Msp I and Sac I at the apo AI-CIII gene loci in a rural Japanese population, H-Y district, Shiso County, Hyogo Prefecture, Japan.
The other aim is to estimate the association between the genotype and serum TC, LDL-C, HDL-C, triglycerides (TG), apo B, apo Al and apo CIII levels in the population.

Study District
H-Y district is located in the north-western part of Hyogo Prefecture, Japan. Most of the residents are salaried employees of minor companies, who supplement their income with part-time jobs in forestry and agriculture.

Subjects
All residents of H-Y district aged 40 years or older, 2,510 men and 2,621 women, were invited to undergo an examination in 1992, which was conducted under a Japanese law for the prevention of cardiovascular disease. Of these, 73% (1,832/2,510) of the men and 76% (1,992/2,621) of the women responded to the invitation. As for the persons aged 20-39 years, only those who desired to be examined were recruited.
Not only to genotype four RFLPs but also to measure serum lipids and apolipoproteins, a total of 893 unrelated subjects were randomly selected from the respondents. The sample size was determined, taking into account the statistical power to detect the significant difference in the allele frequencies between the Japanese and other populations. Individuals who refused to be genotyped, who took lipid-lowering medication, and who suffered from diabetes mellitus, chronic renal and hepatic diseases and chronic alcohol abuse were excluded. Finally, 673 subjects were recruited for the present analyses.
We obtained informed consent from all of the subjects. Ethical approval for the present study was granted by the Ethical Review Committee, Medical Research Institute, Tokyo Medical and Dental University. The gel samples were stained with ethidium bromide. The presence or absence of cutting site were designated to ' +', or '-' respectively. Different 20 samples were measured twice for each of Xba I, EcoR I, Sac I and Msp I polymorphisms to assess the reproducibility, which was found to be 100%.
The allele frequency was estimated by the gene-counting method10)

Statistical Analyses
SAS software (Version 6.08) was used for all statistical analyses11).
The frequency distributions of values for serum lipids and apolipoproteins were examined whether they followed the normal distribution or not. Since, of them, serum TG and apo CIII concentrations appeared to be approximately log-normally distributed, the values were logarithmically transformed for statistical analyses, although the nontransformed means are presented in the tables.
ANOVA was performed using the general linear model procedure (PROC GLM) to estimate the association between each of the RFLPs and each of serum lipids or apolipoproteins. Age was included as an independent variable in the model to adjust its effect. Least-Square means, the values for genotype class means after adjustment for age, were also calculated.
Since persons homozygous for the X+ allele (X+X+) of Xba I and E-allele (E-E-) of EcoR I were very few in H-Y district, they were also combined with those heterozygous for the corresponding allele, X + X -or E+ E-, for ANOVAs.
In addition, Spearman correlation coefficients were calculated between RFLPs and serum traits, adjusting age effect, in order to estimate the strength of association.

Serum Lipids and Apolipoproteins
The means and standard deviations of serum lipids and apolipoproteins are presented in Table 2. The mean of serum apo B-related trait (TC, LDL-C or apo B) increased with age, reaching the maximum at the age group of 50-59 in men and at the age group of 60-69 in women. Then, afterwards, it decreased. Women showed lower levels than men until the age of 49 years, but their levels surpassed men's at the age of 50 years and over.
The mean or median of serum TG showed the same change with age as that of serum TC in both sexes.
No age-dependent change was observed in the mean of serum HDL-C, while the levels in women were higher than those in men. The mean of serum apo Al had a peak at the age group of 50-59 in both men and women, while women showed higher means than men after the age of 50 years. The mean or median of serum apo CIII tended to decrease with age in men, but to increase in women.

Allele and Genotype Frequencies
The gender-and age-specific allele frequencies of four RFLPs for apo B and AI-CIII genes are shown in Table 3. The rare allele frequencies of Xba I RFLP were in the range of 0.020 to 0.051 in men and 0.014 to 0.038 in women : 0.031-0.122 and 0.050-0.153 for EcoR I, 0.277-  Table 4. Association of genotypes with serum lipids (mmol/L) or apolipoprotein (mg/dL) according to the ANOVAs, age-adjusted.   As for all of the four RFLPs, the genotype frequency distribution in each age group by sex was close to or not significantly different from the Hardy-Weinberg prediction (all of x2<6.00, d.f.=2, p>0.05)

Associations of RFLPs with Serum Lipids and Apolipoproteins
The results of the ANOVA are shown in Table 4. Xba I RFLP was not associated with variation in any biochemical trait, but suggestively (0.05* p < 0.10) associated with serum HDL-C and apo AI in women. In women, the genotype for EcoRIRFLP was significantly associated with serum TC, LDL-C and apo B. The genotype for Sac I RFLP was significantly associated with serum TG and suggestively associated with serum HDL-C and apo CIII in men and with serum LDL-C and apo CIII in women. Msp I RFLP was significantly related to serum apo CIII and suggestively to serum apo B in men. (Another ANOVA was done for the association of Xba I or EcoR I RFLPs with serum traits, in which the minor and heterozygous genotypes were combined. As presented in Appendix 1, the results were almost similar to those of non-combined analyses, Table 4).
As shown in Table 5, Spearman correlation coefficients were relatively small between RFLPs and serum traits with the absolute value less than 0.2.

Table6.
Allele frequencies of RFLPs at the apo B and Al-CIII gene loci among different populations (reviews).

DISCUSSION
The levels of serum total cholesterol for both men and women in H-Y district are almost the same as those of the whole of Japan according to the Japanese National Survey of Circulatory Disorders, 199012). For women, the average values of serum HDL-C in H-Y district were 0.078 mmol/L (3 mg/dL) lower than those in whole Japan, although there was no difference for men. Therefore, the subjects of the present study did not appear to differ markedly from the average Japanese.
The rare allele frequencies of Xba I (X +) and EcoR I RFLPs (E-) in H-Y district ( Table 3) were much lower than those in American and European populations ( Table  6) According to the Japanese National Nutrition Survey29), fat intake was 20-30 g/day and the P/S ratio of the diet (P : polyunsaturated fat, S : saturated fat) was 1.0-1.5 during the national privation period before 1960. Fat intake increased during the high economic growth period from 1960 to 1975, reached the level of 55-60 g/day in 1975, and was stabilized during the low economic growth period from 1975 to 1990. Although the level of fat intake is still much lower and the P/S ratio is much higher in current Japanese than in Europeans and Americans30-32> there is no marked difference in the average level of serum cholesterol between Japan12) and the US33) recently. Thus, from an ecological point of view, this fact might be partially explained by genetic susceptibility of serum cholesterol to dietary fat exposures. Xba I and EcoR I RFLPs, of which allele frequencies are extremely different between these two populations, are likely to be related to this kind of genetic susceptibility. There are some dietary manipulation studies which support this hypothesis34-38) The frequencies of some genotypes, e.g. X-X-which might be associated with the high sensitivity to high saturated fatty acids and cholesterol diet37), were extremely high in the Japanese population.
Sac I and Msp I RFLPs were significantly or suggestively associated with serum TG, HDL-C, apo B, LDL-C or apo CIII in men or women. Some studies suggested an association of Sac I with TG positively23,56-58) and with HDL-C inversely57.58), but others did not59). The persons with M+M+ genotype of Msp I RFLP tended to have higher level of serum HDL-C or apo Al than those with M-M-genotype57,60) (Appendix 2). Thus, we can not conclude that the four RFLPs selected in the present study are genetically strong factors for the levels of serum lipids and apolipoproteins, although the statistical associations of some RFLPs with serum traits were observed.