ABCA1 gene variants regulate postprandial lipid metabolism in healthy men

Abstract

Objective

Genetic variants of ABCA1, an ATP-binding cassette (ABC) transporter, have been linked to altered atherosclerosis progression and fasting lipid concentration, mainly high density lipoproteins (HDL) and Apolipoprotein A1 (APOA1), but results from different studies have been inconsistent.

Methods and results

In order to further characterize the effects of ABCA1 variants in human postprandial lipid metabolism, we studied the influence of three single nucleotide polymorphisms (SNPs) [i27943 (rs2575875); i48168 (rs4149272); R219K (rs2230806)] in the postprandial lipemia of 88 normolipidemic young men, who were given a fatty meal. For i27943 and i48168 SNPs, fasting and postprandial values of APOA1 were higher, and postprandial lipemia was much lower in homozygotes for the major alleles, for total triglycerides in plasma, and large-triglyceride rich lipoproteins (TRL) triglycerides. These persons also showed higher APOA1/APOB ratio. Major allele homozygotes for i48168 and i27943 showed additionally higher HDL and lower postprandial Apolipoprotein B (ApoB).

Conclusions

Our work shows that major allele homozygotes for ABCA1 SNPs i27943 and i48168 have a lower postprandial response as compared to minor allele carriers. This finding may further characterize the role of ABCA1 in lipid metabolism.

ATP-binding cassette (ABC) transporters are a family of proteins that act as trans-membrane carriers of molecules, using ATP hydrolysis as energy source¹. ABCA1, a member of the ABC family, has been implicated in monocyte differentiation and phagocyte/dendritic cell commitment², but the most studied aspect regarding ABCA1 is its regulation of lipid metabolism. ABCA1 is a major regulator of HDL metabolism². Moreover, mutations in the ABCA1 gene are the underlying mechanism of Tangier's disease, an affliction defined by a stark HDL deficiency, mildly to moderately increased triglyceride levels and decreased LDL cholesterol^3-5. Based on “in vitro” and “in vivo” studies, a broader importance in lipid metabolism has been attributed to ABCA1, which now is known to regulate the migration of lipid molecules through the cell membrane ^6-8. Its expression is highly influenced by intracellular cholesterol changes⁹, pro-oxidant substances¹⁰ and lipid loading (via LXR receptors)^{10, 11}. The human ABCA1 gene maps to chromosome 9q31, and its 50 exons span over 150 kbp^{12, 13}. Although the exact physiological effects of ABCA1 are not fully understood, a role in atherosclerosis progression via reverse cholesterol pathway, has been proposed.

A common strategy to indirectly define ABCA1 function has relied on assessing the clinical phenotype associated with variations in its gene. One of the most studied ABCA1 variants is R219K (rs2230806). The status of minor allele carriers has been linked to reduced atherosclerosis, with a decrease in the intima media thickness of the carotid artery^{14, 15}, less severe coronary artery disease (CAD) and slower CAD progression^{14, 16}. Other gene variations also have been studied, with variable effects on lipid concentrations or atherosclerosis ^{14, 17, 18}. We recently reported the interaction of ABCA1 with ABCG5/ABCG8 gene variants on HDL concentration¹⁹. However, and paradoxically, with increasing evidence for an effect of these variants on atherosclerosis and CAD comes contradiction regarding their effects on fasting lipid concentrations¹⁴. Although a typically plausible underlying mechanism of this altered atherosclerosis was the changes in HDL concentration, this has not been found in the majority of studies^{14, 15}. Looking for additional physiological pathways underlying ABCA1 effects on lipid metabolism and atherosclerosis, we investigated and report here the effects of ABCA1 variants i27943, i48168 and R219K on postprandial lipid metabolism of healthy males.

Methods

Subjects

Eighty-eight healthy men aged 18 to 33 years were selected among students from the University of Cordoba. We included only young normolipemic APOE E3/E3 males in order to avoid possible effects of different APOE isoforms or gender. Other results of this cohort have been published elsewhere^20-22. No participants had diabetes or liver, renal or thyroid disease, nor were they taking any medication. Anthropometric measures (weight, height, and body mass index) and blood pressure were assessed and all subjects were encouraged to maintain regular lifestyle and levels of physical activity. All volunteers had plasma cholesterol and triacylglycerol concentrations below 200 mg/dl. Baseline characteristics of the participants are summarized in Table 1. The study in which these participants were enrolled was approved by the Ethics Committee for Clinical Investigations of the Reina Sofía University Hospital in Cordoba, and participants previously signed an informed consent to join the study.

Table 1.

Baseline characteristics of the study participants.

		CHOL (mg/dl)	TG (mg/dl)	HDL (mg/dl)	LDL (mg/dl)	APOA1 (mg/dl)	APOB (mg/dl)
	GG n=50	153.61 ± 3.39	86.7 ± 4.6	45.64 ± 1.38	90.68 ± 3.1	103.64 ± 2.55	67.88 ± 2.48
ABCA1 R219K	GA/AA n=34/4	150.48 ± 3.61	72.90 ± 5.5	47.01 ± 1.83	88.84 ± 3.46	109.36 ± 3.58	67.72 ± 2.53
(rs2230806)	p	0.537	0.056	0.545	0.695	0.184	0.965
	CC n=23	152.54 ± 4.78	73.64 ± 7.29	50.061 ± 2.07*	88.05 ± 4.07	116.43 ± 4.04*	63.50 ± 3.13
ABCA1 i48168	CT n=51	152.82 ± 3.06	80.76 ± 4.67	45.36 ± 1.32	90.72 ± 3.2	103.59 ± 2.59	69.16 ± 2.00
(rs4149272)	TT n=14	149.33 ± 5.85	81.56 ± 8.93	45.389 ± 2.53	87.4 ± 5.38	106.75 ± 4.95	67.30 ± 3.83
	p	0.867	0.685	0.041	0.820	0.033	0.310
	GG n=21	152.12 ± 5.00	75.63 ± 7.67	49.36 ± 2.21	85.36 ± 5.74	115.78 ± 4.28†	63.88 ± 3.29
ABCA1 i27943	GA n=52	153.347 ± 2.99	79.88 ± 4.59	45.89 ± 1.32	91.39 ± 3.02	104.00 ± 2.56	69.06 ± 1.97
(rs2575875)	AA n=15	147.86 ± 5.83	81.17 ± 8.93	46.12 ± 2.57	87.46 ± 4.44	107.89 ± 4.98	66.38 ± 3.82
	p	0.707	0.867	0.395	0.574	0.039	0.385

The p-value in each cell corresponds to univariate ANOVA, with each genotype as independent factor and each phenotype variable as dependent factor (with age and BMI as covariates). Within each cell, the upper genotype corresponds to homozygotes for the major allele, the intermediate to heterozygotes, and the lower to homozygotes for the minor allele. All values are mean ±SE. Superscripts in cells are as follows:

^*=p<0.05 ABCA1 i48168 CC vs CT.

^†=p<0.05 ABCA1 i27943 GG vs GA

Study design

After an overnight, 12-h fast, subjects were given a fatty meal enriched with 60,000 units of Vitamin A per m² body surface area. The amount of fat given was 1 g of fat and 7 mg of cholesterol per kg body weight. This meal contained 60% of its energy in the form of fat (35% SAT, 19% MUFA, 6.3% PUFA), 15% as protein and 25% as carbohydrate, and was consumed within 20 min. After the meal, subjects were not allowed another energy intake for 11 h, but were permitted to drink water. Blood samples were drawn just prior to the meal, and postprandially every hour until 6 h, then every 2 h and 30 min until 11 h. Taking samples at such late time-points allowed many lipid measures to return to near fasting levels.

Biochemical determinations

SNP selection, DNA amplification and genotyping

R219K G>A is a well characterized SNP, which has been extensively studied, and associated with CVD, but its influence on postprandial lipemia has not been tested¹⁴. We have reported previously on other effects of SNPs i27943G>A and i48168G>A¹⁹. Computational analysis ascribed potential functional characteristics to each variant allele of both SNPs¹⁹. Additionally, for the i48168 G> A polymorphism, analysis by MAPPER indicated a potential allele-specific binding site for the cartilage paired-class homeoprotein 1 (CART1 or ALX1) transcription factor, whose motif appears enriched in certain genes involved in cholesterol metabolism (Parnell LD, Ordovas JM, unpublished data). Finally, both SNPs showed an informative allele frequencies in reference populations. Based on these premises, we selected these three SNPs as good potential candidates. SNPs were genotyped using the Applied Biosystems TaqMan assay ^23-25. Allele discrimination was performed on PCR products. Fluorescence data were collected by a 7900 Sequence Detection System (Applied Biosystems)²³.

Lipoprotein separation and lipid analysis

Large and small TRL were manually extracted after centrifugation in subdued light as previously described and samples were stored at -70°C until analyzed ²⁵. Total cholesterol (Chol) and triglycerides (TG) in plasma and lipoprotein fractions were assayed by enzymatic procedures^{26, 27}. APOA1 and APOB were determined by turbidimetry ²⁸. HDL-C was measured by analyzing the supernatant obtained following precipitation of a plasma aliquot with dextran sulfate-Mg²⁺, as described²⁹. LDL-C levels were estimated using the Friedewald formula, based on the CHOL, TG, and HDL-C values ³⁰.

Statistical analysis

Statistical analysis methods employed here are similar to those which we have published previously with respect to the gene-postprandial state interaction^{31, 32}.

Genotype analysis

Linkage disequilibrium was tested using Helix-Tree software (Helix-Tree, version 4.3.2. Golden Helix, Bozeman, MT, USA). Likelihood ratio test was used to determine the existence of linkage disequilibrium. When linkage disequilibrium was observed, r² was used to measure its strength. Using r² values, we classified linkage disequilibrium as weak (<0.30), moderate (0.30-0.80) or strong (>0.80). Hardy-Weinberg equilibrium (HWE) was tested by Fisher's exact test.

Analysis of lipid parameters

The influence of the SNPs on the size of the postprandial lipid fractions was analyzed by one-way ANOVA for area under the curve (AUC), defined as the area between the plasma concentration-versus-time curve, using the trapezoidal rule, with the SNPs included as independent factors, and BMI and age as covariates. For any lipid fraction studied, a linear regression model was constructed to determine the influence of the covariates on the dependent variables. LSD or Games-Howell adjustments were used, depending on Levene's test for homogeneity of variances. If the variances were homogeneous, the method for correction was LSD adjustment, while non-homogeneous variances called for applying the Games-Howell method. Repeated-measures ANOVA was also used for both the overall gene influence (global ANOVA, p for gene influence), the kinetics of the response (p for time), and the interaction of both factors (time*gene). When statistical differences were found in the repeated measured ANOVA, a multiple comparisons test adjusted by LSD was applied to identify differences among the genetic isoforms on each time-point of extraction. An alternative approach to the postprandial hypertriglycidemia was done by the accumulated TG evolution: Concentration of TGs were added to each previous timepoint to acquire each “timepoint accumulated TGs”. For example, “timepoint accumulated TG” at the third hour was computed as the fasting TG concentration plus the concentrations of TG at timepoints 1, 2 and 3.

When minor allele homozygotes were less than 5% or a dominant effect of the allele was inferred, we also studied the stratified data of the mutant allele carriers versus the homozygotes for the common allele. A p-value of less than 0.05 was considered significant. All data presented in the text and tables are expressed as mean ± S.E unless otherwise specified. SPSS 17.0 for Windows (SPSS Inc., Chicago, IL, USA) was used for statistical comparisons.

Results

Characteristics of participants at fasting state and genotype frequencies are shown in Table 1. APOA1 was higher in homozygotes for the common allele of i27943 and i48168 vs heterozygotes for both SNPs. HDL was higher in homozygotes for the common allele of i48168 vs heterozygotes. For the three SNPs analyzed, there was no departure from Hardy-Weinberg (p>0.05). Pairwise linkage disequilibrium in correlation coefficients of the three SNPs were as follows: SNPs i27943 and i48168 were in strong linkage disequilibrium: p<0.05; r²=0.827. R219K was not in linkage disequilibrium with either of the other two SNPs (r²<0.012, p>0.05). The influence of the SNPs on postprandial lipid levels is described below, and is summarized in Table 2, and Figures 1 through 3.

Table 2.

Area under the curve of lipid fractions in the postprandial study (mean ± SE).

	ABCA1 R219K	ABCA1 i48168	ABCA1 i27943
Total TG	GG 79.5 ± 5.0	CC 56.6±7.3*	GG 56.8 ± 7.8‡
	GA/AA 70.9 ± 5.9	CT 76.5±4.7	GA 76.4 ± 4.7
		TT 89.6±9.0	AA 87.1 ± 9.1
CHOL	GG 80.9 ± 1.7	CC 79.9±2.7	GG 78.5 ± 2.8
	GA/AA 80.8 ± 2.1	CT 81.1±1.7	GA 81.9 ± 1.7
		TT 79.9±2.7	AA 79.0 ± 3.3
large-TRL TG	GG 33.4 ± 2.5	CC 20.9±3.5*	GG 21.6 ± 3.7‡
	GA/AA 28.7 ± 3.0	CT 31.8±2.3	GA 31.3 ± 2.3
		TT 38.2±4.4	AA 37.3 ± 4.5
small-TRL TG	GG 23.7 ± 2.0	CC 20.3±3.1	GG 20.1 ± 3.2
	GA/AA 23.4 ± 2.4	CT 23.9±2.0	GA 24.1 ± 2.0
		TT 24.9±3.9	AA 24.2 ± 3.9
large-TRL CHOL	GG 4.7 ± 0.3	CC 4.7±0.4	GG 4.8 ± 0.4
	GA/AA 4.8 ± 0.3	CT 4.7±0.2	GA 4.7 ± 0.3
		TT 4.6±0.5	AA 4.7 ± 0.5
small-TRL CHOL	GG 6.3 ± 0.4	CC 5.5±0.7	GG 5.7 ± 0.7
	GA/AA 6.0 ± 0.5	CT 6.1±0.5	GA 5.9 ± 0.4
		TT 6.5±0.9	AA 6.9 ± 0.9
APOA1	GG 54.0 ± 1.3	CC 60.1±1.9†	GG 59.1 ± 2.1§
	GA/AA 56.6 ± 1.5	CT 53.6±1.2	GA 54.2 ± 1.2
		TT 55.9±2.4	AA 55.9 ± 2.4
APOB	GG 35.1 ± 1.1	CC 32.6±1.7	GG 31.8 ± 1.7§
	GA/AA 35.5 ± 1.3	CT 36.2±1.1	GA 36.5 ± 1.0
		TT 34.8±2.1	AA 34.2 ± 2.0
HDL	GG 23.7 ± 0.8	CC 26.0±1.2	GG 25.3 ± 1.3
	GA/AA 24.2 ± 0.9	CT 23.5±0.7	GA 23.9 ± 0.8
		TT 23.8±1.4	AA 23.8 ± 1.5

Univariate ANOVA using BMI and age as covariates. TG: Triacylglycerols; Chol: Cholesterol; TRL: TG Rich Lipoproteins. All values are expressed as (min*mg/dl)/10³. Superscripts in cells are as follows:

^*=p<0.05 ABCA1 i48168 CC vs CT and CC vs TT;

^†=p<0.05 ABCA1 i48168 CC vs CT;

^‡=p<0.05 ABCA1 i27943 GG vs GA and GG vs AA;

^§=p<0.05 ABCA1 i27943 GG vs GA.

Figure 1.

Evolution of TG and Large-TRL TG concentrations depending on ABCA1 i48168 (Panels A and B) and i27943 (panels C and D) genotype. * p<0,05 i48168 CC vs TT; ** p<0,05 i48168 CC vs CT; ^† p<0,05 i27943 GG vs AA; ^‡ p<0,05 i27943 GG vs GA.

Figure 3.

Evolution of APOA1 concentration (mg/dl) during the postprandial test, depending on ABCA1 i48168 genotype (panel A) and i27943 (panel B) genotype. * p<0,05 i48168 CC vs CT; **p<0,05 i27943 GG vs GA.^† <0,05 i27943 GG vs AA.

R219K

A statistical study is presented for a genotype dominant effect based on previously published data (for review, see Iatan, et al ¹⁴). In parallel an additive model was also performed, but we did not observe any differences compared with the dominant model. A trend for lower fasting TG and large-TRL TG was found in minor allele carriers compared to major allele homozygotes (p=0.056 and p=0.070 respectively, Table 2). We did not find other significant differences in the postprandial lipid metabolism.

i48168

CC individuals of i48168 (homozygotes for the common allele) showed a lower AUC of total TG compared to the other two genotypes (p=0.006 vs TT and p=0.025 vs CT, Table 2). In the repeated measures ANOVA, we found lower TGs for CC vs TT from hours 1 to 8.5, and vs CT from the 2^nd to 5^th time points (Figure 1A). CC participants had lower AUC of large-TRL TGs than the other two groups (Table 2). The differences were noted at timepoints 3 to 8.5 vs TT, and 2 to 6 vs CT (Figure 1B).

Subjects homozygous for the major allele displayed lower amounts of accumulated TG from the third hour to the end of the study, compared to CT and TT (all p<0.05) (Figure 2A). We found no differences in the postprandial state for total cholesterol, large-TRL Chol, small-TRL Chol or small-TRL TG depending on this SNP. A trend for higher AUC of HDL was noted for CC homozygotes vs heterozygotes (p=0.074). In the repeated measures ANOVA, the differences were significant at fasting in CC vs CT (p=0.034). AUC of APOA1 was higher in CC subjects than in CT (p=0.006). Differences between CC and CT were significant at all timepoints, from fasting to the 11^th hour (Figure 3A). Although the trend was similar for CC vs TT, significance was not achieved at any timepoint (p ranging from 0.07 to 0.45). The APOA1/APOB ratio was higher in CC versus TC individuals (p=0.008).

Figure 2.

Accumulated TG concentration (mg/dl) during the postprandial test, depending on ABCA1 i48168 genotype (panel A) and i27943 (panel B) genotype. * p<0,05 i48168 CC vs TT; ** p<0,05 i48168 CC vs CT; ^† p<0,05 i27943 GG vs AA; ^‡ p<0,05 i27943 GG vs GA.

i27943

Major allele homozygotes for i27943 (GG) had a lower AUC of TG than GA and AA subjects (Table 2). In the repeated measures ANOVA, differences were significant at timepoints 2 to 8.5 vs AA and 2 to 6 vs GA (Figure 1C). AUC of large TRL-TG was lower for GG participants than for GA and AA (Table 2). Differences were noted at timepoints of 3 to 8.5 hours in GG and AA subjects, and only at the 3^rd hour between GG and GA (although p-values were lower than 0.10 at timepoints 2, 5 and 6) (Figure 1D). We did not find any effects of this SNP on small-TRL TGs, total cholesterol, large-TRL chol, small-TRL chol, or HDL. AUC of APOA1 was higher for GG vs GA (Table 2). In the post-hoc analysis we found differences between GG and GA at fasting and hours 1, 4, 6 and 11 after the meal (with additional p<0.10 at timepoints 2, 3, 5 and 8.5), and between GG and AA at timepoint 8.5 (Figure 3B). A lower AUC of APOB was observed for GG vs GA subjects (Table 2). In the repeated measures ANOVA, there was lower APOB in GG vs GA at all timepoints (all p<0.05) except at fasting, and hour 11. The ratio of APOA1 to APOB was higher in GG versus GA.

Discussion

Persons homozygous for the major alleles of ABCA1 SNPs i27943 and i48168 have lower postprandial lipemia than carriers of minor alleles, in healthy young men. These results derive from a highly controlled, standardized trial of APOE E3/E3 participants, who were subjected to a lipemia test meal. The interaction between genes and postprandial lipemia is well reported³³. The complexity of conducting a study with hourly blood draws and a total duration of 11 hours is high, and, hence, this type of study is rarely conducted and cannot be performed in larger and broader epidemiological studies due to methodological issues. Although other simpler designs for postprandial lipemia assessment have been reported, our method allows deep evaluation into the postprandial state³⁴.

The relationship between ABCA1 polymorphisms and altered atherosclerosis is well stated ^{14, 17, 35}. To date, the strongest associations have linked ABCA1 variants with altered HDL concentrations ^{14, 18, 36}, which are probably mediated by APOA1 metabolism¹⁵. However, these variations in HDL have not been stated universally. Furthermore, in a recent report in type 2 diabetic patients, the ABCA1 SNPs associated with CHD (including R219K) were not associated with HDL, and those associated with HDL were not associated with CHD³⁷. The explanation for these contradictory findings has been set on the limited effects that gene variation can have on final HDL levels, gene-environment interactions, or the influence of ABCA1 gene variants on other lipid molecules and enzymes which secondarily can mildly influence HDL concentrations¹⁷.

The minor allele of R219K has been associated with limited atherosclerosis¹⁵, reduced risk for myocardial infarction or progression of coronary disease in various studies ^{16, 38-44}. Here, we noticed only a trend toward lower fasting TG (p=0.059), according to previous studies ¹⁴. Currently, this variant is thought to affect lipids mildly, but gene-environment interactions are strong, with greater effects on lipid concentration when oxidative stress or inflammation is elevated in subjects (14, 15, 41, 42). Such is not the case in our study. However, marginal effects, not reaching significance, were found in our study for practically all lipid parameters towards a protective effect for the minor allele of R219K, which could become significant when the subject is exposed to the above mentioned stressors, or even simply with increasing age.

Because SNPs i27943 and i48168 are in strong linkage disequilibrium in our population, it is possible that part of the results obtained for one SNP may be due to linkage disequilibrium and not to real functional effects. If this were the case, we hypothesize that i48168 has a greater likelihood to be the functional SNP based in three points. First, i48168 has been shown to influence lipids in which i27943 showed no effects¹⁹. Second, computational analysis indicated a potential allele-specific binding site for the cartilage paired-class homoprotein 1 (CART1) transcription factor, whose motif appears enriched in certain genes involved in cholesterol metabolism¹⁹. Third, the significance coefficient (p value) was lower when the two SNPs were observed to influence the same lipid parameters. Although part of the effects may be secondary to their high linkage disequilibrium, other findings support that both variants have differential effects. In this sense, i27943 alone showed association to altered plasma APOB levels in the postprandial state.

Our results show that homozyogtes for the common alleles of i27943 and i48168 had lower postprandial increases in TG and large-TRL TG, indicating a lower lipemic response, and lower fasting and postprandial APOA1 levels. The effects on TG (TG and large-TRL TG) were observed with an “additive-model” in the sense that heterozygotes were more similar to homozygotes for the major allele, and minor allele homozygotes showed the highest postprandial lipemia. Furthermore, the accumulated TG in the postprandial state showed time-dependent divergent lines for each genotype, which manifests as an accumulation of TG (clearance delay) during the postprandial period. In HDL, we only noted a higher fasting value for i48168 common allele homozygotes versus heterozygotes. Major allele homozygotes for i27943 showed lower APOB concentrations in the postprandial state, but curiously, not in the fasting state. In both cases, homozygotes for the major alleles displayed a higher APOA1/APOB ratio than heterozygotes, which has been associated with cardiovascular risk⁴⁵.

There is a lack of studies describing effects of these two SNPs on human lipid metabolism. A recent study from our group in a cohort of Puerto Rican participants (aged 45-70 years), living in Boston (USA), reported an interaction of i48168 with ABCG5 and ABCG8 variants, to determine HDL levels¹⁹. ABCG5/ABCG8 SNPs were associated with differences in HDL concentration only in minor allele homozygotes. The lower HDL in carriers of the minor allele found in our study can support the theory that these persons have some deregulation of HDL metabolism that leads to different HDL values, just as we observed, or to a higher susceptibility to altered HDL concentration when exposed to other gene variants, as in the Puerto Rican study.

To the best of our knowledge, this is the first report of postprandial data on these two ABCA1 SNPs, probably because HDL metabolism has been the primary focus of research in humans. Nevertheless, animal models have repeatedly shown effects on the postprandial state, although the influence of loss of ABCA1 function has been linked to both increased and decreased postprandial triglycerides^{46, 47}. Furthermore, and supporting postprandial effects of these proteins in humans, patients with Tangier's disease follow a pattern similar to the one that we observed, with delayed TG clearance⁴⁸. In a recent review of metabolic regulation of intestinally derived lipids, these apparent contradictory findings were noted, with the authors stating that “clearly, either positively or negatively, ABCA1 appears to influence postprandial lipid metabolism”⁴⁸.

Further study is clearly required to unify the classic model, in which effects of ABCA1 polymorphisms were found mainly in APOA1 and HDL concentrations, with our results, in which most of the effects appear to be produced in the postprandial state. This is probably best accomplished by in vitro/in vivo studies, which will require a focus on the postprandial clearance of particles in models with these SNPs. In our point of view, APOA1 may be the cornerstone of the effects found with SNPs i48168 and i27943. As is broadly known, APOA1 protein is mainly present in HDL-cholesterol. However, it is also present on the surface of nascent chylomicrons. We have reported recently that an APOA1 SNP [APOA1 -2803; (rs2727784)] clearly influences postprandial lipemia in healthy males²⁰. Furthermore, effects of this variant on postprandial lipemia were quite similar to those we report here, showing a clear influence on the molecules that initiate postprandial metabolism, such as total and large-TRL TGs (which express APOA1 on their surface), and no effect on molecules implicated in the final phase of postprandial metabolism (mainly small-TRL), which do not express APOA1 in their surface. Our explanation for the effects of APOA1 -2803 variant, which can also be applied to the ABCA1 SNPs reported here, stated that effects of the variant on postprandial metabolism may be mediated by altered levels of APOA1 present on the surface of large TRL. Supporting this theory, we have found alterations in total APOA1 concentrations depending on the ABCA1 variants i48168 and i27943.

In conclusion, in our study, healthy young men carrying the minor alleles for i48168 and i27943 of ABCA1 show much higher postprandial lipemia. This feature has been associated with a higher risk for developing accelerated progression of atherosclerosis, and eventually cardiovascular disease, but identification of phenotype association in a clinical trial to increase cardiovascular risk can be overlapping, and requires further investigation. Extrapolation to other age groups, or to people with associated conditions, however, may not be correct, as it has been reported that the effects of other variants of ABCA1 on lipid metabolism are highly dependent on those factors.