A Novel Fatty Acid Lipophilic Index and Risk of Coronary Heart Disease in U.S. Men: the Health Professionals Follow-Up Study

Abstract

Few epidemiologic studies have examined the association between an overall fatty acid (FA) profile and coronary heart disease (CHD) risk. To examine a novel index that summarizes individual FA levels based on FA affinity and fluidity in relation to CHD risk in men. In a prospective nested case-control study, FAs in plasma and erythrocytes were measured in 459 CHD cases and 879 matched controls. Lipophilic index (LI) was computed by summing the products between FA levels and melting point of each FA to reflect the overall FA lipophilicity. Among controls, higher plasma LI was significantly correlated with adverse profiles of blood lipids, inflammatory markers, and adiponectin. After multivariate adjustment for age, smoking, body mass index, and other CHD risk factors, plasma LI was significantly associated with an increased risk of CHD: the relative risk was 1.61 (95% confidence interval: 1.03, 2.53; P for trend=0.04) comparing extreme quintiles. This association was attenuated to 1.21 (0.48, 3.09; P for trend = 0.77) after adjusting for plasma levels of total trans FAs, long-chain n-3 FAs, and polyunsaturated to saturated fat ratio. Erythrocyte LI was not significantly associated with CHD risk. Our data indicate that a novel lipophilic index is associated with an adverse profile of cardiovascular risk markers and increased risk of CHD in men, its usefulness as a complement of individual fatty acids in assessing disease risk needs to be elucidated in future studies.

INTRODUCTION

Fatty acids are a group of molecules sharing a common structure of hydrophilic carboxyl terminal and a hydrophobic hydrocarbon chain that jointly endow fatty acids the amphipathic nature. Because of this nature, fatty acids are essential components of phospholipids, which are building blocks of cellular or blood lipid membranes. Fatty acid composition in membranes determines the fluidity of the membranes, which subsequently influences the functions of membrane-bound proteins and cells⁽¹⁾. Specifically, membrane fluidity is largely dictated by the lipophilic attraction among fatty acids in phospholipid layers, which depends primarily on the length of hydrocarbon chain and number of double bond of fatty acids. Fatty acids from both diet and de novo synthesis are incorporated into the membrane phospholipids and subsequently modify the membrane’s viscosity as well as its functions⁽²⁾.

Multiple lines of evidence have suggested that the membrane fluidity of cells and lipoproteins may play a role in the etiology of coronary heart disease (CHD) through various pathways, such as effects on blood pressure, blood lipid metabolism, and endothelial function^(3–5). Epidemiologic data regarding membrane fluidity and CHD risk, however, are lacking because the current broad fatty acid classifications do not take into account their diverse biological properties such as affinity and fluidity. A unified approach to summarizing an overall fatty acid profile that reflects membrane fluidity is needed to examine the hypothesis of interest.

We, therefore, developed a novel index⁽⁶⁾ to assess fatty acid lipophilicity by summarizing levels of individual fatty acids and their melting points, which measure the lipophilic attraction between fatty acids⁽⁷⁾. In the current analysis, we aimed to prospectively evaluate this novel lipophilic index (LI) of fatty acids in plasma and erythrocyte membranes in relation to CHD risk among U.S. men in a prospective nested case-control study within the Health Professionals Follow-up Study (HPFS) cohort.

RESEARCH DESIGN AND METHODS

Study Population

The HPFS is an ongoing prospective cohort study consisting of 51,529 U.S. male health professionals who were 40–75 years old at study inception in 1986. Medical history, lifestyle practices, and diet were assessed at baseline and updated every 2–4 years using self-administered questionnaires since study baseline. In 1993–1995, a total of 18,159 participants provided blood samples, which were centrifuged and aliquoted into cryotubes as plasma, buffy coat, and erythrocytes upon arrival. All cryotubes were stored in the vapor phase of liquid nitrogen freezers at a temperature ≤ −130°C. A nested case-control study of CHD was conducted among these participants who provided blood samples⁽⁸⁾. Briefly, among those who were free of cardiovascular disease at blood draw, we prospectively identified incident CHD cases and selected one to two controls for each case using the risk-set sampling method from those who remained free of CHD events when the case was diagnosed. Cases and controls were matched on age (± 2 years), smoking status (never smoke, past smoker, current smoker: 1–14 cigarettes/day, 15+ cigarettes/day), and month of blood draw. Through 2008, a total of 460 CHD cases including 358 cases of nonfatal myocardial infarction (MI) and 102 cases of fatal CHD were identified and confirmed, and 894 controls were selected. All analyses were conducted for plasma and erythrocyte LIs separately. A total of 14 cases and 56 controls had missing plasma fatty acid levels were excluded in plasma LI analyses; while 2 case and 18 controls had missing erythrocyte fatty acids were excluded from erythrocyte LI analyses. After these exclusions, 446 CHD cases and 838 controls were included for plasma LI analyses, and 458 CHD cases and 876 controls were included for erythrocyte LI analyses. Among these participants, a total of 787 men provided fasting blood sample (time since last meal before blood draw ≥ 8 h). There were no significant differences of plasma LI and erythrocyte LI values between fasting samples and non-fasting samples (18.46 vs. 18.24 for plasma LI; 25.64 vs. 25.67 for erythrocyte LI).

The study protocol was approved by the institutional review board of Brigham and Women’s Hospital and the Human Subjects Committee Review Board of Harvard School of Public Health.

Assessment of CHD endpoint

Study physicians who were blinded to exposure status reviewed medical records of participants who reported having MI in follow-up questionnaires. Nonfatal MI cases were confirmed using the World Health Organization criteria, which require typical symptoms plus either electrocardiographic abnormality or elevated cardiac enzyme levels⁽⁹⁾. Fatal CHD was identified through reports from next of kin, from postal authorities, or by searching the National Death Index. Fatal CHD defined as per ICD-9 codes 410–412 and was confirmed by reviewing hospital records or autopsy reports, if CHD was listed as the cause of death on the death certificate and if evidence of previous CHD was available in the medical records. For cases with CHD as the underlying cause on the death certificate but no medical records concerning the death were available and no prior knowledge of CHD was indicated, we designated such cases as probable fatal CHD cases⁽⁸⁾. We excluded disconfirmed CHD deaths in the case-control study. Because the exclusion of probable fatal CHD events (n=9) did not alter the results, we included both confirmed and probable fatal CHD in this analysis to maximize statistical power. Total CHD was defined as nonfatal MI plus fatal CHD.

Fatty acids measurement and LI calculation

Fatty acids of total plasma and erythrocyte membranes were measured by gas-liquid chromatography, which has been described in detail elsewhere⁽¹⁰⁾. Briefly, fatty acids were extracted from plasma and erythrocyte membranes using a hexane-isopropanol mixture and esterified with methanol and acetyl chloride. After esterification, the methanol and acetyl chloride were evaporated, and the fatty acid methyl esters were re-dissolved in isooctane. The methyl esters were analyzed using gas-liquid chromatography. Peak retention times and area percentages of total fatty acids were identified by injecting known standards (Nu-Chek-Prep, Elysium, MN). A total of 35 plasma fatty acids and 36 erythrocyte membrane fatty acids were identified. The content of each fatty acid was expressed as a percentage of total fatty acids.

Samples of matched case-control sets were handled identically and assayed in the same analytical run. Both technicians and laboratory personnel were blinded to case-control status of the samples. Laboratory control samples were run along with case-control samples. Coefficients of variation (CVs) of the assay were assessed by repeatedly analyzing quality-control samples. The average intra-assay CV was 13% for both plasma and erythrocyte fatty acids with relative content higher than 0.2%. For those fatty acids with relative content lower than 0.2%, the average CV was 22% and 30% for plasma and erythrocyte fatty acids, respectively. The intra-assay CVs for individual fatty acids were shown in Supplementary Table 1.

The LIs for plasma and erythrocyte fatty acid composition were calculated as a summation of the product of the levels of fatty acids (percentage of total fatty acid) and the melting point⁽¹¹⁾ of each fatty acid (in °C) using the following equation:

$$\text{lipophilicindex}=\frac{{\displaystyle \sum _{\kappa }}[\text{Levels}\text{of}\text{fatty}\text{acid}{(\%)}_i\times \text{melting}\text{point}(°\text{C})i]}{{\displaystyle \sum _{\kappa }}[\text{Levels}\text{of}\text{fatty}\text{acid}{(\%)}_i]}$$

where _i=individual fatty acid, k=number of fatty acids used to calculate LI. This index reflects the overall fatty acid lipophilicity in plasma or erythrocytes.

In the current study, all fatty acids with available melting points (n=26), including 14:0, 15:0, 16:0, 17:0, 18:0, 19:0, 20:0, 22:0, 23:0, 24:0, 16:1n-7, 18:1n-7, 18:1n-9, 20:1n-9, 24:1n-9, 18:2n-6, 18:3n-6, 20:4n-6, 18:3n-3 (α-linolenic acid; ALA), 20:5n-3 (eicosapentaenoic acid; EPA), 22:6n-3 (docosahexaenoic acid; DHA), trans-16:1n7, total trans 18:1 isomers, and three trans 18:2 isomers (9t,12t 18:2n-6, 9c,12t 18:2n-6 and 9t,12c 18:2n-6) were used to compute fatty acid LI in plasma and erythrocytes. Because individual trans 18:1 isomers were not separated, we used the mean melting point of trans 18:1n-12, trans 18:1n-9 and trans 18:1n-7 as melting point of total trans 18:1. The melting points for selected fatty acids range from 52.3 to 87.8 °C for saturated fatty acids (SFA), 0 to 42.8 °C for monounsaturated fatty acids (MUFA), −54.1 to −5.0 °C for polyunsaturated fatty acids (PUFA), and 1.0 to 48.7 °C for trans fatty acids (TFA). These fatty acids covered the majority of total fatty acids in plasma (96%) and erythrocyte membranes (92%).

Measurement of plasma lipids and inflammatory markers

Concentrations of total cholesterol, triglycerides, and high-density lipoprotein (HDL) cholesterol were measured on the Hitachi 911 analyzer using reagents and calibrators from Roche Diagnostics (Indianapolis, IN); all CVs were <1.8%. Low-density lipoprotein (LDL) cholesterol levels were assessed using a homogenous direct method from Genzyme (Cambridge, MA); CV was <3.1%. C-reactive protein (CRP) was measured by latex-enhanced immunoturbidimetric assay from Denka Seiken (Tokyo, Japan) on the Hitachi 911 system, and CV was <2.8%. Soluble tumor necrosis factor (TNF)-α receptor 1 and 2 (TNFα-R1 and TNFα-R2) and interleukin-6 (IL-6) levels were measured using enzyme-linked immunosorbent assays from R&D Systems (Minneapolis, MN) with CVs of 3.5–9.0%⁽¹²⁾. Intercellular adhesion molecule-1 (ICAM-1) and vascular cell adhesion molecule-1 (VCAM-1) were measured using a commercial enzyme-linked immunosorbent assay (R & D Systems, Minneapolis, MN); CVs were <3.6%. Plasma adiponectin concentrations were measured by competitive radioimmunoassay (Linco Research, St. Charles, MO) with a CV of 3.4%⁽¹³⁾.

Assessment of covariates

Information about weight, smoking status, aspirin use, cholesterol-lowering medication use, and physical activity was inquired about in the baseline questionnaire. Body mass index (BMI) was calculated as weight/height² (kg/m²). Physical activity was expressed as metabolic equivalents per week using the duration of moderate or vigorous forms of exercise multiplied by the intensity of the activity⁽¹⁴⁾. History of hypertension, hypercholesterolemia, and diabetes and family history of MI were based on self-reports. Total energy intake and consumptions of carbohydrate, fats, and alcohol were estimated using a validated food frequency questionnaire. Alternate Healthy Eating Index (AHEI) score was calculated based on intakes of fruit, vegetable, nuts, trans fat, the ratio of polyunsaturated to saturated fatty acids (P:S), and cereal fiber⁽¹⁵⁾.

Statistical Analysis

Among controls, linear regression analysis was applied to examine the associations between individual fatty acids and LI. We calculated partial Spearman correlation coefficients (r), which were adjusted for matching factors to evaluate the correlations between LI and blood lipids, inflammatory markers, and other cardiovascular risk factors. We used conditional logistic regression to examine the association between LI and risk of CHD. Participants were categorized into quintiles based on the distribution of LI among the controls. Besides matching factors, we further controlled for physical activity (in quartiles), alcohol intake (none, 0.1–4.9, 5.0–9.9, 10.0–14.9, 15.0+ g/day), BMI (continuous), family history of MI (yes/no), history of hypertension (yes/no), history of hypercholesterolemia (yes/no), aspirin use (yes/no), cholesterol-lowering medication use (yes/no), hypertension medication use (yes/no), and AHEI score. Tests for linear trends were conducted by treating the median value for each quintile of LI as a continuous variable.

All P values were 2-sided, and 95% confidence intervals (CIs) were calculated for Relative Risks (RRs). Data were analyzed with the Statistical Analysis Systems software package, version 9.1 (SAS Institute, Inc., Cary, NC, US).

RESULTS

The association between LIs and individual fatty acids among controls are shown in Table 1. Both plasma and erythrocyte LIs were positively associated with individual SFAs, including 14:0, 15:0, 16:0, 17:0, 18:0, 19:0, 20:0, 22:0, 23:0, and 24:0 (all βs ≥ 0.17, P < 0.001). In contrast, LIs were inversely correlated with all PUFAs, including ALA, EPA, DHA, 18:2n-6, 18:3n-6, and 20:4n-6 (all βs ≤ −0.25, P < 0.001). Both plasma and erythrocyte LIs showed inverse association with MUFA, including 16:1n-7, 18:1n-7, and 18:1n-9 (all βs ≤ −0.03, P < 0.001), except 24:1n-9 (β = 0.26 for plasma LI and 0.18 for erythrocyte LI, P < 0.001). The correlations with plasma and erythrocyte LIs were not consistent among trans fatty acids. For example, both plasma and erythrocyte LIs were positively correlated with total trans 18:1 isomers (β= 0.31 for plasma LI and 0.26 for erythrocyte LI, P < 0.001), whereas inverse correlations were observed for 9c,12t 18:2n-6 and 9t,12c 18:2n-6 isomers (β = −0.20 and −0.18 for plasma LI and −0.35 and −0.16 for erythrocyte LI, P < 0.001). Plasma LI, but not erythrocyte LI, was also significantly correlated with levels of plasma trans 16:1n-7 (β = 0.10 for plasma LI, P < 0.001). Using the forward selection procedure of linear regression, we found that the strongest determinants of plasma LI were 16:0, total trans 18:1 isomers, 18:0, 20:4n-6, 18:1n-9, and DHA in plasma, whereas the major determinants of erythrocyte LI were 18:0, 14:0, DHA, 18:2n-6, 20:4n-6 and 18:1n-9 in erythrocyte membranes. The correlation coefficient between plasma LI and erythrocyte LI was 0.13 (P < 0.001).

Table 1.

Multiple linear regression analysis of selected fatty acids in relation to lipophilic index, the Health Professionals Follow-up Study.^a

Variable	Melting Point	Plasma lipophilic index (n=838)				Erythrocyte lipophilic index (n=876)
		Mean	SD	Beta	SE	Mean	SD	Beta	SE
Saturated fatty acids
14:0	53.9	0.53	0.38	0.37	0.004	0.30	0.22	0.17	0.018
15:0	52.3	0.14	0.05	0.30	0.025	0.11	0.06	0.50	0.043
16:0	63.1	19.1	2.46	0.46	0.002	19.2	1.97	0.41	0.002
17:0	61.3	0.31	0.09	0.44	0.012	0.37	0.10	0.39	0.041
18:0	69.6	8.17	1.33	0.52	0.002	14.5	2.26	0.46	0.003
19:0	68.6	0.10	0.05	0.50	0.022	0.12	0.04	0.37	0.035
20:0	76.8	0.19	0.05	0.59	0.025	0.38	0.05	0.69	0.039
22:0	81.5	0.49	0.19	0.66	0.009	1.58	0.28	0.63	0.011
23:0	79.1	0.20	0.08	0.61	0.024	0.28	0.06	0.31	0.05
24:0	87.8	0.42	0.16	0.71	0.012	3.90	0.81	0.68	0.005
Monounsaturated fatty acids
16:1n-7	0.0	1.55	0.79	−0.20	0.003	0.46	0.23	−0.24	0.012
18:1n-7	15.0	1.65	0.29	−0.04	0.005	1.08	0.14	−0.18	0.016
18:1n-9	16.0	19.4	3.01	−0.03	0.002	13.1	2.07	−0.1	0.002
20:1n-9	23.3	0.15	0.07	−0.01NS	0.017	0.19	0.04	−0.05 NS	0.047
24:1n-9	42.8	0.54	0.23	0.26	0.007	3.77	0.80	0.18	0.004
Polyunsaturated fatty acids
18:2n-6	−5.0	30.6	4.45	−0.25	0.002	13.2	3.07	−0.33	0.002
18:3n-6	−11.2	0.42	0.15	−0.31	0.008	0.1	0.04	−0.54	0.042
20:4n-6	−49.5	7.29	1.78	−0.71	0.002	13.0	1.82	−0.81	0.003
18:3n-3	−11.2	0.60	0.24	−0.31	0.004	0.21	0.24	−0.37	0.008
20:5n-3	−54.1	0.65	0.48	−0.75	0.003	0.52	0.31	−0.84	0.007
22:6n-3	−44.2	1.75	0.77	−0.65	0.002	3.7	1.22	−0.76	0.002
Trans fatty acids
trans 16:1n-7	31.0	0.15	0.06	0.11	0.018	0.14	0.05	0.02 NS	0.05
trans 18:1	48.7	1.74	1.01	0.31	0.002	1.48	0.64	0.26	0.004
9t, 12t 18:2n-6	28.5	0.04	0.03	0.22	0.027	0.01	0.02	−0.02 NS	0.075
9c, 12t 18:2n-6	1.0	0.24	0.11	−0.20	0.01	0.11	0.04	−0.35	0.083
9t, 12c 18:2n-6	1.0	0.16	0.11	−0.18	0.01	0.08	0.04	−0.16	0.084

^NSnon-significant.

^aAll P<0.001 unless otherwise indicated.

Associations between baseline lifestyle and dietary factors and plasma or erythrocyte LIs among control participants are shown in Table 2. Both plasma and erythrocyte LIs were consistently correlated with less favorable risk factors of CHD, including higher BMI and lower levels of physical activity and AHEI score. In addition, participants with higher plasma LI levels were more likely to be current smokers and have a history of hypertension and hypercholesterolemia. Erythrocyte LI levels were not correlated with these factors. Both plasma and erythrocyte LIs were weakly associated with higher intakes of SFA (r = 0.16, P<0.001 for both LIs), lower intakes of PUFA (r = −0.14 for plasma LI and r = −0.16 for erythrocyte LI, P <0.001), and lower dietary P:S ratio (r = −0.15 for plasma LI and r = −0.17 for erythrocyte LI, P <0.001). We did not observe significant association between LIs and intakes of MUFAs or TFAs. We observed moderate correlations between dietary LI and LIs of plasma and erythrocyte: the correlation coefficient was 0.16 for plasma LI and 0.17 for erythrocyte LI, respectively (P < 0.001 for both correlations).

Table 2.

Baseline characteristics of control participants by quintiles of lipophilic index at baseline in 1994, the Health Professionals Follow-up Study^a

Characteristics	Plasma lipophilic index						Erythrocyte lipophilic index
	Q1		Q3		Q5		Q1		Q3		Q5
	Mean	SD	Mean	SD	Mean	SD	Mean	SD	Mean	SD	Mean	SD
No. of participants	167.0		168.0		167.0		175.0		176.0		175.0
Plasma lipophilic index	14.6	1.0	17.9	0.4	22.5c	2.2	17.4	2.7	18.5	2.7	18.5c	3.0
Erythrocyte lipophilic index	25.5	3.3	25.7	2.9	25.8	2.0	22.8	1.2	25.2	0.3	29.7 c	3.3
Demographic characteristics
Age, years b	63.3	8.7	63.3	8.8	65.0	7.9	63.6	8.9	63.8	8.5	63.2	8.8
BMI, kg/m2	24.7	2.3	25.4	2.8	26.2c	3.2	25.0	2.9	25.8	2.7	25.7 c	3.4
Physical activity, MET-hours/week	23.2	26.9	19.1	22.5	15.9	17.4	22.8	43.2	21.5	19.4	17.6	20.9
Current smoker, % b	4.3		8.9		15.2c		6.0		12.4		9.9
History of hypertension, %	22.8		23.2		40.7c		28.0		25.6		31.4
History of high cholesterol, %	38.3		38.7		47.3 d		48.6		37.5		38.9d
Family history of MI, %	31.7		29.8		37.1		31.4		34.7		37.1
Hypertension medication user, %	18.6		14.3		31.1 c		22.3		18.8		20.0
Cholesterol-lowering medication user, %	6.6		6.6		5.4		9.1		4.6		8.0
Aspirin user, %	26.4		19.1		24.6		25.7		20.5		21.7
Diet
Total energy intake, kJ/d	8501.5	2446.0	8519.0	2684.0	8696.0	2924.2	8332.0	2332.2	8548.3	2568.1	8570.9	2912.9
Total carbohydrate intake, g/d	246.2	38.0	242.2	39.6	244.7	41.9	249.1	37.7	244.0	42.5	240.1 c	38.8
Glycemic load	130.5	22.6	128.1	23.2	129.5	25.2	131.0	21.7	128.7	24.9	126.7 c	22.6
Total fat intake, g/d	68.5	12.0	70.0	12.7	68.2	13.2	66.6	12.3	68.4	12.6	72.1 c	12.5
Saturated fat intake, % total fat	32.1	3.6	33.6	3.6	34.0c	3.8	32.4	3.6	33.2	3.8	34.2 c	3.4
Monounsaturated fat intake, % total fat	39.1	1.8	39.1	2.2	38.8	2.1	38.9	2.1	38.9	2.0	38.9	2.0
Polyunsaturated fat intake, % total fat	19.7	3.2	18.2	3.1	18.1c	3.3	19.6	3.3	18.8	3.3	17.9 c	2.9
Trans fat intake, % total fat	4.3	1.1	4.2	1.2	4.5	1.3	4.2	1.3	4.2	1.2	4.4	1.1
P:S ratio	0.54	0.20	0.46	0.17	0.46c	0.16	0.53	0.19	0.49	0.19	0.44c	0.14
Alcohol, g/d	11.8	13.0	12.9	13.8	13.5	16.5	13.6	15.2	14.5	16.2	10.9	14.0
AHEI diet score	43.9	8.9	41.6	9.5	41.0c	9.4	43.9	9.8	42.3	9.5	40.0 c	7.8

Abbreviations: MI, myocardial infarction; AHEI, alternative healthy eating index; P:S, polyunsaturated to saturated fatty acids.

^aData are mean and SD for continuous variables or % for categorical variables, unless indicated otherwise.

^bAge and smoking status are matching factors.

^cP for trend <0.01.

^dP for trend <0.05.

Supplementary Table 2 presents partial Spearman correlations of plasma and erythrocyte LIs with blood lipids and other CHD risk markers among controls. After adjusting for matching factors including age, smoking, and month of blood draw, higher plasma LI was significantly correlated with adverse profiles of CHD risk factors, including elevated levels of triglycerides (r = 0.58, P < 0.001), total:HDL cholesterol ratio (r = 0.33, P < 0.001), LDL:HDL cholesterol ratio (r = 0.14, P < 0.001), CRP (r = 0.10, P = 0.003), IL-6 (r = 0.18, P < 0.001), TNFα-R1 (r = 0.18, P < 0.001), TNFα-R2 (r = 0.13, P = 0.005), ICAM-1 (r = 0.10, P = 0.02), and VCAM-1 (r = 0.12, P = 0.008), and lower levels of HDL cholesterol (r = −0.37, P < 0.001) and adiponectin (r = −0.21, P < 0.001). These correlations were much weaker for erythrocyte LI. Both plasma and erythrocyte LIs were inversely correlated with LDL cholesterol (r = −0.18 and −0.11, respectively, P < 0.002).

The baseline characteristics of cases and controls are presented in Supplementary Table 3. The associations of plasma and erythrocyte LIs with risk of CHD are shown in Table 3. In a crude model that controlled for matching factors only, the RR was 1.76 (95% CI: 1.15, 2.69; P for trend = 0.01) comparing the highest to the lowest quintile of plasma LI. After multivariable adjustment of CHD risk factors, including BMI and lifestyle practices, the RR comparing extreme quintiles was attenuated to 1.61 (95% CI: 1.03, 2.53; P for trend = 0.04). In a sub-group analysis, we found that the association for plasma LI was more pronounced for nonfatal MI (RR = 1.81, 95% CI 1.09, 3.01; P for trend = 0.01) in comparison to that for fatal CHD (RR = 1.05, 95% CI 0.33, 3.33; P for trend = 0.67). Although a positive trend was also observed for erythrocyte LI, the association did not achieve statistical significance.

Table 3.

Relative risks (95% confidence intervals) of coronary heart disease across the quintiles of plasma and erythrocyte lipophilic index, the Health Professionals Follow-up Study.^a

	Q1	Q2	Q3	Q4	Q5	P for trend
Plasma lipophilic index
Median (range)	15.0 (10.9–15.8)	16.5 (15.9–17.2)	17.9 (17.3–18.6)	19.4 (18.7–20.3)	21.7 (20.4–33.4)
Total CHD
Case/control	74/167	81/168	83/168	95/168	113/167
Model 1b	1.00	1.27 (0.84, 1.91)	1.22 (0.81, 1.85)	1.42 (0.94, 2.12)	1.76 (1.15, 2.69)	0.01
Model 2c	1.00	1.23 (0.80, 1.88)	1.25 (0.81, 1.93)	1.34 (0.88, 2.06)	1.61 (1.03, 2.53)	0.04
Non-fatal MI
Case/control	60/130	63/133	65/124	75/124	81/120
Model 1b	1.00	1.16 (0.74, 1.82)	1.28 (0.81, 2.02)	1.51 (0.96, 2.37)	1.78 (1.10, 2.86)	0.01
Model 2c	1.00	1.19 (0.74, 1.90)	1.40 (0.86, 2.27)	1.52 (0.94, 2.45)	1.81 (1.09, 3.01)	0.01
Fatal CHD
Case/control	14/26	18/21	18/31	20/38	32/34
Model 1b	1.00	1.89 (0.72, 4.97)	1.05 (0.39, 2.82)	1.14 (0.46, 2.84)	1.68 (0.65, 4.35)	0.53
Model 2c	1.00	1.54 (0.48, 4.90)	0.85 (0.27, 2.65)	0.72 (0.24, 2.13)	1.05 (0.33, 3.33)	0.67
Erythrocyte lipophilic index
Median (range)	23.1 (17.0–23.8)	24.3 (23.9–24.8)	25.2 (24.9–25.6)	26.2 (25.7–26.9)	28.3 (27.0–42.0)
Total CHD
Case/control	78/175	97/175	95/176	100/175	88/175
Model 1b	1.00	1.37 (0.91, 2.08)	1.41 (0.91, 2.18)	1.58 (1.01, 2.49)	1.45 (0.88, 2.37)	0.22
Model 2c	1.00	1.42 (0.92, 2.19)	1.58 (1.00, 2.50)	1.69 (1.06, 2.70)	1.35 (0.80, 2.26)	0.41
Non-fatal MI
Case/control	65/142	74/134	71/129	72/133	74/124
Model 1b	1.00	1.22 (0.78, 1.92)	1.25 (0.77, 2.01)	1.39 (0.86, 2.27)	1.47 (0.87, 2.49)	0.17
Model 2c	1.00	1.28 (0.80, 2.05)	1.43 (0.87, 2.35)	1.51 (0.91, 2.51)	1.44 (0.83, 2.51)	0.24
Fatal CHD
Case/control	13/28	23/27	24/39	28/30	14/32
Model 1b	1.00	2.65 (0.88, 7.97)	2.86 (0.87, 9.39)	3.39 (0.95, 12.1)	1.39 (0.32, 6.0)	0.92
Model 2c	1.00	3.04 (0.77, 12.02)	3.75 (0.85, 16.64)	3.68 (0.78, 17.48)	1.02 (0.16, 6.40)	0.46

Abbreviations: CHD, coronary heart disease; MI, myocardial infarction.

^aThe relative risks (RRs) were estimated using conditional logistic regression. Miss values for plasma and erythrocyte lipophilic index were n=54 and n=4, respectively.

^bModel 1: adjusted for matching factors included age (years), smoking status (never/current smoker/ past smoker) and month of blood draw.

^cModel 2: multivariate RR additional adjusted for alcohol intake (g/d: 0, 0.1–4.9, 5.0–9.9, 10.0–14.9, and >15), physical activity (quartile), family history of MI (yes/no), history of hypertension (yes/no), history of hypercholesterolemia (yes/no), aspirin use (yes/no), cholesterol-lowering medication use (yes/no), antihypertensive medication use (yes/no), alternative healthy eating index and BMI.

In a secondary analysis, we explored whether the association between plasma LI and CHD was independent of plasma levels of fatty acids that are established risk factors of CHD, such as P:S ratio, long-chain n-3 fatty acids (EPA and DHA), and total trans fatty acids. The multivariable-adjusted RRs (95% CI) comparing extreme quintile of plasma levels of P:S ratio, long-chain n-3 fatty acids, and total trans fatty acids were 0.61 (0.40, 0.95, P for trend = 0.04), 0.64 (0.42, 1.00, P for trend = 0.04) and 1.45 (0.94, 2.23, P for trend = 0.14), respectively. After adjusting for these three aforementioned factors, the RR (95% CI) comparing extreme quintiles of plasma LI was attenuated to 1.21 (0.48, 3.09; P for trend = 0.77). In a separate analysis, we evaluated whether the association of plasma LI was driven by fatty acids that were the strongest predictors of plasma LI identified in the forward selection analysis, i.e., 16:0, total trans 18:1 isomers, 18:0, 20:4n-6, 18:1n-9 and DHA. The RRs (95% CIs) of total CHD comparing extreme quintiles were 1.42 (0.77, 2.60), 1.63 (1.02, 2.61), 1.67 (1.07, 2.64), 1.30 (0.75, 2.27), 1.20 (0.71, 2.05), and 1.44 (0.89, 2.35) after further adjustment for 16:0, total trans 18:1 isomers, 18:0, 20:4n-6, 18:1n-9, and DHA, respectively. The association was substantially attenuated when all six plasma fatty acids were adjusted simultaneously; the RR (95% CI) was 1.04 (0.37, 2.94) comparing extreme quintiles. Lastly, we examined whether plasma lipids and inflammatory markers explained the associations of interest. Further adjustment for triglycerides and HDL cholesterol attenuated the association between plasma LI and CHD: the RR (95% CI) of total CHD comparing extreme quintiles was 1.11 (0.67, 1.86; P for trend = 0.77). Adjustment for CRP, IL-6, TNFα-R1, TNFα-R2, ICAM-1, or VCAM-1 somewhat strengthened the association. The RRs (95% CIs) of total CHD comparing extreme quintiles were 1.59 (1.01, 2.49) with CRP adjustment, 1.90 (1.04, 3.48) with IL-6 adjustment, 1.90 (1.04, 3.47) with TNFα-R1 adjustment, 1.89 (1.03, 3.47) with TNFα-R2 adjustment, 1.88 (1.02, 3.44) with ICAM-1 adjustment, and 1.84 (1.01, 3.37) with VCAM-1 adjustment.

DISCUSSION

In this nested case-control study among U.S. men, we examined a novel fatty acid lipophilic index that represents overall lipophilicity of fatty acids in plasma or erythrocyte membranes in relation to CHD risk. Plasma LI was significantly associated with higher risk of developing CHD, especially nonfatal MI. Further adjustment for fatty acids that were established risk factors of CHD and blood lipids substantially attenuated the association between plasma LI and CHD. The associations were weaker for erythrocyte LI.

It is well-known that membrane lipid contents determine membrane fluidity or miscibility and, subsequently, the membrane’s physiological function⁽¹⁶⁾. The inter-fatty acids attraction, one of the main factors regulating the membrane fluidity, is determined by the van der Waals interaction, which depends on two main molecular characteristics of fatty acids: the length of the fatty acid hydrocarbon chain and fatty acid unsaturation, i.e., number of double bonds⁽¹⁶⁾. With the exception of trans fatty acids which have overall linear structure because of the trans configuration of double bonds, longer hydrocarbon chain and more saturation will lead to tighter molecular packing or more lipophilic attraction. Such lipophilic attraction is reliably measured by the melting point of fatty acids⁽⁷⁾. This is the basis of using melting points to derive our lipophilic index, which theoretically reflects the relative fluidity of cellular or lipoprotein membranes.

Membrane fluidity may influence CHD risk through multiple pathways. Changes of membrane fluidity would modulate the activity of proteins involved in ion transport, signal transduction, cell Ca2+ handling, and intracellular pH regulation⁽¹⁷⁾, suggesting that decreased membrane fluidity may play an important role in the pathogenesis of hypertension^{(4; 18; 19)}. In addition, reduced membrane fluidity could also significantly alter the vascular endothelial response to shear stress and impair endothelial cell wound closure⁽⁵⁾. Moreover, low membrane fluidity may interfere the sodium-dependent D-glucose transport^{(20; 21)}. There is also evidence indicating that membrane fluidity is strongly related to insulin sensitivity^{(20; 22)}. In fact, effects of Metformin in diabetes treatment may be partially via its effects on membrane fluidity⁽²³⁾. In addition to cell membrane fluidity, the fluidity of lipoprotein membranes may also play a critical role in CHD etiology. Fluidity of lipoproteins phospholipid layer is known to be affected by phospholipid fatty acyl composition of lipoproteins⁽²⁴⁾. Phospholipid fluidity of HDL particles regulates the activity of lecithin:cholesterol acyltransferase⁽²⁵⁾ and the capacity of HDL to promote cholesterol efflux⁽²⁶⁾.

Of note, in the current study we only calculated lipophilic indices of total plasma and erythrocyte membranes, the most accessible tissues in large epidemiological studies involving hundreds to thousands of human participants. We found a positive association between plasma LI and CHD risk. Although plasma fatty acid content may not directly measure that of cell membrane involved in the aforementioned mechanisms, evidence suggests that plasma phospholipid compartment actively exchanges phospholipids with plasma membranes of cells^{(27; 28)}. Therefore, plasma phospholipids may serve as a dynamic and readily available pool for cell membrane phospholipids and subsequently affect the membrane fatty acid composition, resulting in change of membrane fluidity. The null association observed for erythrocyte membrane LI was unexpected. Although apparently more data are needed to confirm these findings, several reasons may explain the null results. Erythrocytes are incapable of de novo phospholipid synthesis, and the major pathway for the renewal of erythrocyte phospholipids is direct exchange of phosphatidylcholine with surrounding plasma lipoproteins⁽²⁷⁾. Therefore, in comparison to plasma, erythrocyte membrane phospholipids may serve as a secondary pool for exchanging phospholipids with other cells. In addition, erythrocyte fatty acid measurements had relatively high measurement error (CV was, on average, 27% higher than plasma fatty acids, which might be due to more difficulties in lipid extraction from erythrocytes than plasma. In prospective studies, such random measurement error will, in general, lead to attenuation of true association.

The possibility of alternative explanations that do not relate to membrane fluidity deserves discussion. In the current study plasma and erythrocyte LIs were significantly correlated with high SFA and low PUFA intakes. In addition, trans 18:1 isomers and DHA, fatty acids that are exclusively or largely from diet, were among the strongest individual fatty acids that determine the levels of plasma LI. Intake of these fatty acids can affect CHD risk through multiple pathways. For example, dietary SFA significantly increased risk of CHD primary through effects on increasing plasma concentration of LDL-cholesterol^{(29; 30)}. In contrast, replacement of SFA with PUFA in the diet significantly reduced LDL-cholesterol concentration and CHD risk⁽³⁰⁾. Dietary PUFA has also been indicated to alter fatty acids composition of tissue cell membrane phospholipids and lead to improved insulin sensitivity^(31–33). In addition, long chain n-3 fatty acid intakes were significantly associated with lower levels of triglycerides and inflammatory biomarkers, as well as improved blood pressure and vascular function⁽³⁴⁾. Dietary consumption of trans fat might also contribute to increased risk of CHD via its adverse effects on blood lipids⁽³⁵⁾. Lastly, it is possible that high SFA and low PUFA diet raises CHD risk through promoting thromboembolism. It has been observed that serum agglutination or creaming occurred after fat emulsions were administered intravenously to critically ill patients⁽³⁶⁾, especially when CRP levels were high^(36–38), and such agglutination might lead to increased risk of developing thromboembolism⁽³⁸⁾.

There are several limitations in the current analysis that should be discussed. First, plasma or erythrocyte LIs may not be a direct measurement of membrane fluidity of various cell types and plasma lipoproteins. In the current study, the association of plasma LI was not independent of some individual fatty acids, such as trans 18:1 isomers and DHA, or blood lipids. Therefore, plasma LI did not provide additional predictive value beyond blood lipids and individual fatty acids that were established risk factors of CHD. More data of direct measurements of cell/lipoprotein membrane LIs are warranted to corroborate our observations. Second, we cannot exclude the possibility that the current findings are due to unmeasured and residual confounding that is intrinsic to any observational studies. Third, we only measured plasma and erythrocyte fatty acid levels at baseline. Because fatty acid contents may fluctuate through change of diet, it is likely this snapshot of fatty acid composition measurement may not represent long-term average values. In general, this type of measurement errors tends to nullify associations. Lastly, the generalizability of our findings is restricted to white male health professionals.

In summary, we derived a novel lipophilic index that summarizes the overall lipophilicity of major fatty acids in plasma and erythrocytes. In a prospective study of male U.S. health professionals, we found a significant association between this index in plasma and an increased risk of CHD, and this association was primarily attributable to several individual fatty acids, including trans 18:1 isomers and DHA. More studies are warranted to corroborate our findings and to elucidate whether the LI possesses values in predicting future CHD beyond those of some individual FAs that are established risk factors of CHD.

Abbreviations

CHD

Coronary heart disease

LI

Lipophilic index

HPFS

Health Professionals Follow-up Study

MI

Myocardial infarction

CV

Coefficient of variation

ALA

α-linolenic acid

EPA

Eicosapentaenoic acid

DHA

Docosahexaenoic acid

HDL

High-density lipoprotein

LDL

Low-density lipoprotein

CRP

C-reactive protein

TNFα-R1

Tumor necrosis factor-alpha receptor 1

IL-6

Interleukin-6

ICAM-1

Intercellular adhesion molecule-1

VCAM-1

Vascular cell adhesion molecule-1

BMI

Body mass index

AHEI

Alternate Healthy Eating Index

P:S

Polyunsaturated to saturated fatty acids

SFA

Saturated fatty acid

PUFA

Polyunsaturated fatty acid

MUFA

Monounsaturated fatty acid

TFA

trans fatty acids