Association between APOE Genotype with Body Composition and Cardiovascular Disease Risk Markers Is Modulated by BMI in Healthy Adults: Findings from the BODYCON Study

Body mass index (BMI) has been suggested to play an important role in the relationship between the APOLIPOPROTEIN (APO)E genotype and cardiovascular disease (CVD) risk. Using data from the BODYCON cross-sectional study (n = 360 adults) we assessed the association between body composition and CVD risk markers according to APOE genotype, with examination of the role of BMI. In this study cohort, the APOE2/E3 group had lower fasting blood lipids than APOE4 carriers and APOE3/E3 group (p ≤ 0.01). After stratifying the group according to BMI, APOE4 carriers in the normal BMI subgroup had a higher lean mass compared with the APOE3/E3 group (p = 0.02) whereas in the overweight/obese subgroup, the android to gynoid percentage fat ratio was lower in APOE4 carriers than APOE3/E3 group (p = 0.04). Fasting lipid concentrations were only different between the APOE2/E3 and other genotype groups within the normal weight BMI subgroup (p ≤ 0.04). This finding was associated with a lower dietary fibre and a higher trans-fat intake compared with APOE4 carriers, and a lower carbohydrate intake relative to the APOE3/E3 group. Our results confirm previous reports that BMI modulates the effect of APOE on CVD risk markers and suggest novel interactions on body composition, with diet a potential modulator of this relationship.


Introduction
The APOLIPOPROTEIN (APO)E gene is one of the most widely studied in relation to cardiovascular disease (CVD) risk due to the association with circulating blood lipids. It encodes the multifunctional apoE apoprotein which represents an important ligand for the receptor-mediated uptake of triacylglycerol (TAG)-rich lipoproteins and their remnants from the circulation [1]. The APOE2, APOE3 and APOE4 alleles have different affinities for the low-density lipoprotein (LDL) receptor which impacts on cholesterol homeostasis and blood lipid profile [2][3][4]. It has been well documented that the APOE gene accounts for 7% of the variance in cholesterol in Caucasians [5]. Although several studies have reported elevated total (TC) and LDL cholesterol (LDL-C) concentrations in APOE4 carriers and lower concentrations in APOE2 carriers compared to the wild-type APOE3/E3 group [6,7], these relationships have not been reported by others [8][9][10][11][12]. These inconsistencies between studies have been attributed to the metabolic status and adiposity of the study populations suggesting that other factors such as body mass index (BMI) may impact on the relationship between the APOE genotype and chronic disease risk [10][11][12].
It is well-known that obesity is an independent risk factor for CVD [13]. Animal studies have shown apoE knock out mice to be protected against obesity [14,15] and suggested a differential effect of the APOE alleles on the ability of the body to store fat, with APOE3 mice having a higher body weight than APOE4 mice on a Western type diet [16][17][18]. Moreover, increased visceral adipose tissue (VAT) accumulation, which is associated with increased CVD risk [19,20], was reported in APOE3 compared to APOE4 mice [17,21,22]. Therefore, a possible explanation for the inconsistent results on association between the APOE genotype and blood cholesterol concentrations may be dependent on adiposity. In agreement, several studies have reported the relationship between APOE and blood lipid risk markers to differ depending on BMI but the mechanisms underlying this association are unclear. Lower TAG concentrations in APOE2 carriers compared with the APOE3/E3 group and APOE4 carriers was reported to be evident only in the UK adults with a normal BMI [23], whereas in Mexican Amerindian population, differences in TC, LDL-C and TAG among APOE4 carriers and APOE3/E3 genotype group were only found in obese subjects (BMI ≥ 30 kg/m 2 ) [24]. However, the limited human studies conducted to date have failed to identify which APOE allele is more prone to obesity and whether an interaction exists between APOE and adiposity on CVD risk markers [23][24][25][26][27]. In addition, the APOE genotype may have an impact on food preferences which can affect body composition. However, the evidence is limited [28].
Therefore, this paper aims to investigate the association between the APOE genotype with body composition and CVD risk markers, with further examination of the role of BMI on this relationship.

The Effect of APOE Genotype on Body Composition Measures and Cardiovascular Disease Risk Markers
The main characteristics for 360 participants (187 female and 168 male) according to the APOE genotype is shown in Table 1 and presented for women and men separately in Supplementary Tables S1 and S2, respectively. The study population had an average age of 42 ± 1 y and BMI of 24.1 ± 0.2 kg/m 2 , and n = 46 participants were APOE2/E3, n = 228 the wild type APOE3/E3 group and n = 81 APOE4 carriers (APOE3/E4 and APOE4/E4). Subjects with the APOE2/E4 genotype (n = 5) were not included in the analysis due to the small sample size and no participants with the APOE2/E2 genotype were identified in the study cohort. The APOE allele distribution was found to be in Hardy-Weinberg equilibrium.
In the group as a whole, fasting TC (p = 0.01), LDL-C (p ≤ 0.01) and non-high density lipoprotein cholesterol (HDL-C) concentrations (p ≤ 0.01) and LDL-C:HDL-C ratio (p = 0.02) in the APOE2/E3 group were on average 9-18% lower compared to APOE4 carriers and 9-16% lower compared with the APOE3/3 group (Table 1). Although genotype was found to have a significant effect on diastolic blood pressure (p = 0.04), differences between the genotype groups was not evident after post hoc analysis. Furthermore, anthropometric and body composition measures were not different between the genotype groups. The habitual dietary intakes of participants are shown in Table 1. Total dietary fibre intake was on average 4 g higher in APOE4 carriers than APOE2/E3 group (p = 0.04), while there was no difference in intake in the APOE3/E3 group compared to the other genotype groups. There was also an influence of APOE on total protein intake, with the APOE3/E3 group consuming 3% of total energy (TE) (approximately 5 g) lower than participants in the APOE2/E3 group (p < 0.01), but the intake in the APOE4 carriers was not different to the other genotype groups. The APOE genotype did not affect total dietary energy or intake of other macronutrients. Abbreviations: AOAC-Association of official analytical chemists, A/G fat % ratio-android to gynoid fat % ratio, BMI-body mass index, CHO-carbohydrate, CRP-C-reactive protein, HC-hip circumference, HDL-Chigh-density lipoprotein cholesterol, LDL-C-low-density lipoprotein cholesterol, MUFA-monounsaturated fatty acids, NEFA-non-esterified fatty acids, PUFA-polyunsaturated fatty acids, SFA-saturated fatty acids, TC-total cholesterol, TAG-triacylglycerol, VAT-visceral adipose tissue, WC-waist circumference.
After stratifying the group according to sex, a few differences were evident on the CVD risk markers and dietary intake according to APOE genotype in men and women.
Genotype had a significant impact on blood lipids, dietary energy intake and total carbohydrate and protein intakes in women whereas in men, the effect was on diastolic blood pressure and NEFA. In women, the LDL-C and non-HDL-C concentrations and LDL-C:HDL-C ratio were significantly lower in APOE2/E3 group compared to the APOE3/E3 group and APOE4 carriers (p < 0.01 for each). Dietary energy intake (p = 0.05) was lower in the APOE2/E3 group compared with APOE4 carriers only and was associated with a significantly lower carbohydrate intake (p = 0.02) compared with the APOE3/E3 group but higher total protein intake (p < 0.01) relative to both the APOE3/E3 and APOE4 carrier groups (Supplementary Table S1). In men, diastolic blood pressure and the fasting NEFA concentration were lower in the APOE4 carriers compared to the APOE3/E3 but not APOE2/E3 group (p < 0.01 for each) (Supplementary Table S2).

Effect of APOE Genotype and BMI on Body Composition Measures and CVD Risk Markers
Significant genotype x BMI interactions were observed for android fat mass (a measure of central adiposity) and for the dietary intakes of total polyunsaturated fatty acids (%TE) and total protein (%TE) (p ≤ 0.03) only. There was no significant impact on other measures of body composition, dietary intakes, or CVD risk markers (Table 1).
To assess the effect of the APOE genotype according to BMI, participants were split into normal weight (BMI ≤ 24.9 kg/m 2 , n = 232) and overweight/obese (BMI ≥ 25 kg/m 2 , n = 128) subgroups. The subject characteristics, body composition and CVD risk markers according to the BMI subgroups are shown in Table 2. In the normal weight BMI subgroup, APOE4 carriers had on average a 3 kg higher lean mass (p = 0.02) and 0.24 kg greater android lean mass (p = 0.01) than the wild-type APOE3/E3 group, while there was no difference in lean mass or android lean mass in the APOE2/E3 group compared to the other genotype groups. Fasting LDL-C (p = 0.01) and non-HDL-C (p = 0.02) concentrations were 17% and 15% lower, respectively in the APOE2/E3 group compared to the APOE4 carriers and 15% and 12% lower compared to the APOE3/E3 group. The LDL-C:HDL-C ratio was also 17% lower in the APOE2/E3 group compared to APOE4 carriers (p = 0.04), with no differences found between the APOE3/E3 group compared to APOE2/E3 and APOE4 carriers. TC concentrations were 9% lower in the APOE2/E3 compared with the APOE3/E3 group (p = 0.04), but not in the APOE4 carriers. In the overweight/obese BMI group, the android to gynoid percentage fat ratio was higher in the APOE3/E3 group compared to APOE4 carriers (p = 0.04), while there was no difference in the APOE2/E3 compared to the other genotype groups. Other body composition measures and CVD risk factors did not differ across the three genotype groups in this BMI subgroup (Table 2).
Habitual dietary intakes are presented in Table 3 according to normal and overweight/obese BMI subgroups. In the normal BMI group, while dietary fibre intake (p = 0.02) was 6 g higher, trans-fat (%TE) (p = 0.05) was 0.15% (5.7 g) lower in APOE4 carriers compared to the APOE2/E3 group, with no differences found compared with the APOE3/E3 group. The participants in the APOE2/E3 group also had a lower dietary carbohydrate (%TE) intake compared to the APOE3/E3 group (p = 0.01) only. Moreover, in the normal BMI subgroup, the APOE2/E3 group had the highest total protein (%TE) intake compared to the APOE3/E3 group (p = 0.01), while the carbohydrate intake of the APOE4 carriers was not different versus the APOE2/E3 and APOE3/E3 groups. Dietary intakes were not different between genotype groups in the overweight/obese BMI group (Table 3). Physical activity levels (steps/day, energy expended performing physical activity per day, and percentage time spent performing sedentary, light, or moderate to vigorous physical activity) were not significantly different according to APOE genotype neither in the whole group or after stratifying according to normal and overweight/obese BMI subgroups (Supplementary Tables S3 and S4).

Discussion
This study examined the association between the APOE genotype with body composition and CVD risk factors and the impact of BMI classification on this relationship. Using data from the impact of physiological and lifestyle factors on body composition (BODYCON) cross-sectional study, we found the APOE genotype to impact on the fasting lipid profile, with differences only evident in participants with a normal BMI and in women. Novel associations between genotype and body composition were observed, with divergent effects of APOE on the android to gynoid percentage fat ratio (an estimate of body fat distribution) and lean body mass within the normal and overweight/obese BMI subgroups.
Several studies have reported associations between the APOE genotype and fasting blood lipid risk markers [29]. In agreement with the previous studies [6,7,30], we also observed TC, LDL-C and non-HDL-C concentrations to be significantly higher in APOE4 carriers and APOE3/E3 group compared to the APOE2/E3 group. However, after dividing the cohort into BMI subgroups, LDL-C and non-HDL-C concentrations were only significantly higher in APOE4 carrier and APOE3/E3 groups compared to the APOE2/E3 group in the normal BMI subgroup. Our findings support those of Kofler et al. [23] who reported the lowest fasting TAG concentration in APOE2 carriers only in participants with normal BMI in the FINGEN study where 312 participants living in the UK were prospectively genotyped for APOE. In agreement with this, Kolovou et al. [31] observed the APOE4 allele to be associated with higher TC levels compared with APOE3 allele in normal-weight coronary heart disease patients based in Greece. Although not widely studied [32,33], we also observed sex-dependent effects on the fasting lipid profile, with differences only evident between genotype groups in the women but not men. Of note, BMI was also significantly lower in the women compared with the men [34]. Therefore, our data shows that the effect of APOE on CVD risk markers may be dependent on their BMI and that the negative metabolic effects of high BMI could mask the effect of the APOE genotype on the fasting lipid profile. For example, our findings show that the detrimental effect of an increased BMI outweighs the positive effect of the APOE2 allele on blood lipid risk markers. Diet and physical activity levels are important modifiable determinants of BMI [35]. It should be noted that APOE4 carriers consumed more dietary fibre and less trans-fat compared to APOE2 carriers. Similarly, studies have shown an impact of the APOE genotype on protein intake. For example, in the Australian Imaging, Biomarkers and Lifestyle study of ageing APOE4 carriers were found to have lower protein intakes than non-APOE4 carriers [19]. Therefore, our study may lend support to the findings that APOE has an impact on food preferences which can affect body composition. Furthermore, our study included healthy subjects with a higher-than-average physical activity level. This may have impacted on the fasting lipid profile observed within the genotype groups as exercise has been shown to favourably affect total cholesterol and TAG levels [36]. Thus, further studies are needed examining the role of APOE genotype on food intake and relationship to BMI.
The effect of APOE genotype on body composition has been investigated in animals and a small number of human studies. Arbones-Mainar et al. [17] reported greater increases in abdominal VAT accumulation and body weight after a high fat western type diet (21%TE fat) in APOE3 mice compared to APOE4. Although our participants consumed on average a high fat diet (37%TE), abdominal VAT was not different between the APOE genotype groups. However, findings from human studies investigating the association between the APOE genotype and body composition are inconsistent. Positive associations between the APOE2 allele with waist circumference and BMI have been reported in 230 Croatian subjects aged 20-85 y and in 4660 Caucasian middle-aged men [37,38]. Another study in 290 children aged 8 years reported lower BMI, trunk fat mass and waist circumference in APOE4 carriers compared to non-APOE4 carriers (APOE3/E3, APOE2/E2 and APOE2/E3) [39]. In contrast, in a case-control study including 198 normal weight healthy and 198 obese Saudi university students, the APOE4 allele was positively associated with BMI in overweight and obese subjects (BMI > 25 kg/m 2 ) [40]. These discrepancies between studies might be influenced by the participants sex, age, ethnicity and/or habitual diet. Therefore, it is important to take these factors into consideration when comparing study findings.
In the current study, we found a genotype x BMI interaction on android fat mass. After stratifying the cohort according to BMI, there were no differences in body weight between APOE genotypes in either the normal or overweight/obese BMI subgroups. However, we observed APOE4 carriers with a normal BMI to have higher lean body mass and android lean mass compared to the wild-type APOE3/E3 group. Therefore, this might provide a possible explanation for the lower body fat and VAT mass accumulation in the APOE4 carriers compared to the APOE3/E3 group in animal studies [16,21]. The mechanisms behind the relationship between the APOE4 allele and increased lean body mass are not clear although animal studies have suggested that adiponectin may play a role. In APOE4 mice, a greater increase in adiponectin levels were observed compared to APOE3 mice on an obesogenic diet [41] and the protective role of adiponectin against muscle loss and muscle growth have been described in some studies [42]. In addition, an association between appendicular lean mass (lean tissue in arms and legs) and circulating adiponectin was reported in postmenopausal women [43]. However, in this study adiponectin concentrations were not different between the genotype groups, therefore this potential mechanism needs to be examined in further studies. Moreover, in the overweight/obese BMI subgroup APOE4 carriers had a lower android to gynoid fat percentage ratio suggesting a difference in body fat distribution compared to the wild-type group which had similar dietary intakes and physical activity levels. These findings are interesting since it is well-known that abdominal obesity is associated with dyslipidaemia [44], and our findings suggest that the APOE3/E3 genotype group had lower LDL-C and non-HDL-C concentrations, but higher android body fat distribution compared to APOE4 carriers. Our finding provide support to those of a previous study which reported that APOE4 mice accumulated less VAT than APOE3 mice after following a high fat diet for 6 months [18]. The authors speculated that endoplasmic reticulum stress is a potential mechanism linking APOE and adiposity. Since apoE4 has a lower protein stability and is abnormally folded in the endoplasmic reticulum, increased endoplasmic reticulum stress in APOE4 carriers may have a negative effect on adipogenesis [18]. Moreover, in a study by Huebbe et al. [16] less weight gain in APOE4 compared to APOE3 mice on high and low-fat diets was observed and the authors reported higher expression of fatty acid-binding protein 4, carnitine palmitoyl transferase 1B and uncoupling protein in APOE4 mice which suggested increased fatty acid oxidation in skeletal muscle in APOE4 mice compared to the APOE3 mice. However, as mice do not usually consume high fat diets it is difficult to translate findings from animal studies to humans. Therefore, further clarification of the association between APOE and body fat distribution measures and the potential mechanisms observed are needed in humans.
The use of a dual energy X-ray absorptiometry (DXA) which is known to be an accurate and precise tool for body composition measurement and includes an estimation of abdominal VAT mass is an important strength of this study. In addition, we included the analysis of a range of outcome measures such as physical activity, dietary intake and CVD risk markers in this cohort. Limitations include the cross-sectional study design, retrospective genotyping and small sample size for some genotype groups, especially during the sub-group analysis according to BMI and sex. Moreover, subjects were not stratified according to the median BMI but as normal and overweight/obese subgroups for translation of our findings to UK public health guidance, with only 36% of this cohort having a BMI > 24.9 kg/m 2 . Finally, it should be noted that using BMI as a marker of adiposity to stratify the group has its own limitations since it cannot distinguish between excess body fat and muscle mass.
In summary, our results indicate an interaction between the APOE genotype and BMI, with higher fasting lipid risk marker concentrations only evident in APOE4 carriers compared to the APOE2/E3 group in participants with a normal BMI and in women. Moreover, differential effects on body fat distribution and composition were observed within the BMI subgroups between the APOE4 carriers and the wild-type APOE3/E3 group, with diet also a potential modulator of this relationship. However, the association between APOE genotype, adiposity, sex, diet and CVD risk markers needs further investigation in humans with prospective genotyping to draw a firm conclusion.

Subjects
A total of 360 healthy men and women aged 18-70 y from the BODYCON study were included in the present analysis. Details of the study design have been described previously [34]. Briefly, participants were recruited from Reading and the surrounding areas and inclusion criteria were BMI 18.5-39.9 kg/m 2 , TC < 7.8 mmol/L, TAG < 2.3 mmol/L, fasting blood glucose < 7.8 mmol/L, haemoglobin > 115 g/L for women and 130 g/L for men. Exclusion criteria were having suffered a myocardial infarction/stroke in the past 12 months, history of diabetes or other endocrine disorders, bowel disease, cholestatic liver disease, pancreatitis, cancer, arthritis or fracture deformity of spine or femur, history of bone related surgeries, radio-opaque implants or implanted medical devices, breastfeeding, being pregnant or planning pregnancy in the next 12 months, being a smoker, being on medication for hyperlipidemia, hypertension, inflammation or hypercoagulation, being on a weight reducing diet and excessive alcohol consumption (<14 units/wk). Female subjects taking oral contraceptives or HRT for at least 3 months were included in the study.

Study Design
The BODYCON study is an observational cross-sectional study conducted in the Hugh Sinclair Unit of Human Nutrition at the University of Reading. The main outcomes of the BODYCON study have been described previously [34]. Briefly, participants attended a single study visit in which a fasting blood sample was collected, and anthropometric measurements were taken. Participants also underwent a DXA scan to assess their total body composition. The NHS and University of Reading Research Ethics Committees both gave a favourable ethical opinion for the conduct of the BODYCON study (NHS reference number: 14/SC/1095 and UREC reference numbers: 17/29 and 13/55). Participants were only included in the current analysis if written consent was obtained for the retrospective genotyping for APOE. The BODYCON study was carried out in accordance with the principles of the Declaration of Helsinki and registered at www.clinicaltrials.gov (accessed on 1 January 2022) (NCT02658539).

Anthropometric Measurements
Anthropometric measures were performed with participants wearing light clothing and no shoes. Height was measured by a stadiometer. Body weight and BMI were determined using a Tanita BC-418 scale (TANITA UK Ltd., Middlesex, UK). Waist and hip circumferences were measured using a non-stretch tape measure. To assess the body composition, DXA scan was performed by trained researchers and described elsewhere [34]. Briefly, prior to the scan participants were required to wear clothes without metal fastenings, buttons or zips, and all metal artefacts were removed. For the total body composition scan, participants lay still on the Lunar iDXA scanner bed with Velcro straps around their knees and ankles. All scans were analysed using enCORE Software, version 15 (GE Medical Systems Ltd, Chalfont St Giles, UK) with the advance software package CoreScan, which also estimates the mass and volume of VAT within the abdomen.

Dietary Intakes and Physical Activity
Habitual dietary intake was assessed using a 4-day weighed diet diary. Dietary data was analysed using DietPlan 7 software (Forestfield, Horsham, UK) and dietary intakes were averaged. Participants with dietary intakes outside of the ranges 500-3500 kcal for women and 800-4000 kcal for men were reported to be under/over reporters (n = 4) and excluded from the dietary analysis [45]. Participants with competition of <3 days of diet diary (n = 1) were also excluded from the dietary analysis. Physical activity levels were measured using a tri-axial accelerometer (Actigraph wGT3X+, Actigraph LLC, Pensacola, FL, USA). Participants were asked to wear the accelerometer directly above the right iliac crest during sleeping and waking hours (except for during water-based activities) for four days, including three week days and one weekend day during the same time that dietary intake was assessed. Device initialization, data processing and analysis were conducted using Actilife Data Analysis Software (Version 6.11.5).

Biochemical Analysis
Fasting blood samples collected into the serum separator and K 3 EDTA blood tubes were centrifuged at 1700× g (3000 rpm) for 15 min at room temperature and 4 • C, respectively before aliquoting into Eppendorf tubes and stored at −20 • C. Fasting lipids (TC, HDL-C, non-esterified fatty acids (NEFA), TAG), glucose and high sensitivity C-reactive protein (hs-CRP)) were quantified in the serum sample by using the ILAB 600 (Werfen (UK) Ltd., Warrington, UK) and RX Daytona Plus (Randox Laboratories Limited, Crumlin, UK) clinical chemistry analysers. The Friedewald equation was used to estimate fasting LDL-C concentrations [46] and non-HDL-C was calculated by subtracting HDL-C from TC. Plasma uric acid was measured using Daytona Plus clinical chemistry analyser (Randox Laboratories Ltd., Crumlin, UK). ELISA kits were used to analyse serum insulin (Dako UK Ltd. Ely, UK and Crystal Chem, Inc., Elk Grove Village, IL, USA) and plasma adiponectin (Quantikine kit, R&D Systems, Europe Ltd, Abingdon, UK.).

DNA Extraction and Genotyping
The buffy coat layer was isolated from the blood sample collected into a 9 mL EDTA blood tube prior to the extraction of DNA using a DNA blood mini kit (Qiagen Ltd., Manchester, UK) according to the manufacturers protocol. DNA samples were genotyped for the single nucleotide polymorphisms (SNP) rs429358 and rs7412 with the use of TaqMan SNP genotyping assays on the QuantStudio 3 real time PCR machine (Life Technologies Limited, Paisley, UK).

Statistical Analysis
Statistical analyses were performed using IBM SPSS Statistics version 25 (SPSS Inc., Chicago, IL, USA). Normality of data was checked using Kolmogrov-Smirnov test and Q-Q plots. Hardy-Weinberg equilibrium was tested by a chi-square test. To assess the effect of the APOE genotype and BMI, a general linear model (ANCOVA) was performed using the study outcome measures as the dependent variable, genotype or BMI as a fixed factor and age and sex as covariates. The effect of APOE according to sex was assessed by a general linear model (ANCOVA) using the study outcome measures as the dependent variable, genotype as a fixed factor and age as a covariate after splitting the data according to sex. To assess the effect of adiposity, a BMI x genotype interaction was added to the model. Participants were then stratified into normal and overweight/obese BMI subgroups and analysed using ANCOVA including age and sex as covariates. If a significant genotype effect was found, pairwise comparisons with a Bonferroni correction were carried out. Results are presented as estimated marginal means ± SE and p ≤ 0.05 was considered significant. Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.