High habitual dietary α-linolenic acid intake is associated with decreased plasma soluble interleukin-6 receptor concentrations in male twins

Abstract

Background: α-Linolenic acid (ALA) is associated with a low risk of cardiovascular disease; however, the underlying mechanism is not completely known.

Objective: The objective was to examine whether habitual dietary ALA intake is associated with plasma concentrations of inflammatory biomarkers after control for shared genetic and common environmental factors.

Design: We cross-sectionally studied 353 middle-aged male twins. Habitual diet was assessed with the Willett food-frequency questionnaire. Fasting plasma concentrations of interleukin-6 (IL-6) and its soluble receptor (sIL-6R), high-sensitivity C-reactive protein (hsCRP), and tumor necrosis factor-α (TNF-α) were measured. Linear mixed-effect regression analysis was used to partition the overall association into within- and between-pair associations.

Results: A 1-g increment in habitual dietary ALA intake was associated with 11.0% lower concentrations of sIL-6R (P = 0.004) but not of IL-6 (P = 0.31), TNF-α (P = 0.16), or hsCRP (P = 0.36) after adjustment for energy intake, nutritional factors, known cardiovascular disease risk factors, and medications. After further control for shared genetic and common environmental factors by comparison of brothers within a twin pair, a twin with a 1-g higher ALA intake was likely to have 10.9% (95% CI: 3.7%, 17.6%; P = 0.004) lower sIL-6R concentrations than his co-twin with a low intake, whereas ALA intake was not significantly associated with plasma concentrations of IL-6, TNF-α, or hsCRP. These results were validated by using 1000 bootstrap samples.

Conclusions: Habitual dietary ALA intake is inversely associated with plasma sIL-6R concentrations independent of shared genetic and common environmental influences. Lowering sIL-6R may be a mechanism underlying the cardioprotective properties of habitual dietary ALA. This study was registered at clinicaltrials.gov as NCT00017836.

INTRODUCTION

Dietary α-linolenic acid (ALA) has been suggested to be protective against cardiovascular diseases (CVDs) (1). ALA, a plant-originated 18-carbon n−3 long-chain polyunsaturated fatty acid (PUFA), is essential for humans. It can be used to synthesize other n−3 PUFAs with ≥20 carbon atoms (n−3 ≥20C-PUFAs), including predominantly eicosapentaenoic acid and docosapentaenoic acid, but very limited amounts of docosahexaenoic acid (2).

The cardioprotective mechanisms of ALA are not completely understood. Inflammation is a pathophysiologic pathway for CVD. ALA could suppress arachidonic acid cascade-related inflammation through a reduction in the production of arachidonic acid from linoleic acid. Interleukin-6 (IL-6) plays a role in arachidonic acid cascade-related inflammation through its receptor (3). IL-6 binds to the IL-6 receptor [either the membrane-bound (mIL-6R) or the soluble receptor (sIL-6R)] to form a complex that leads to the cellular response to IL-6 through the signal transduction of glycoprotein 130 (gp 130) (4). The IL-6 trans-signaling process involving sIL-6R, but not mIL-6R, can be inhibited by soluble glycoprotein 130 (sgp 130) (4). sIL-6R is agonistic for IL-6: some cells do not express mIL-6R and thus the response to IL-6 is only through the IL-6/sIL-6R/gp 130 trans-signaling pathway (4). For cells with mIL-6R, sIL-6R amplifies the cellular response to IL-6 (4). Therefore, sIL-6R may be central in the IL-6 inflammatory action through the arachidonic acid cascade, which could be inhibited by ALA.

Previous studies of the association of dietary ALA with inflammatory biomarkers have yielded inconsistent results (5–10). One possible explanation for these inconsistencies is that genetic and/or common environmental factors may confound the association between habitual dietary ALA and the inflammatory response. Genetic influences are observed for dietary preferences (11) and for circulating concentrations of sIL-6R (12) and IL-6 (13). Because dietary habits in adult life are influenced by those acquired during childhood and youth (14, 15), they are likely to be associated with other environmental conditions shared by members of the same family; therefore, the latter can confound the association between diet and disease. In this context, the study of adult twins reared in the same family is especially useful, because it can help to disentangle specific environmental exposures (such as diet) from shared genes and common environment.

The primary aim of this study was to examine the association between habitual dietary ALA intake and plasma concentrations of IL-6 and sIL-6R. We secondarily tested the association of ALA intake with plasma concentrations of tumor necrosis factor-α (TNF-α), an inflammatory factor that up-regulates IL-6, which, in turn, stimulates the production of C-reactive protein (CRP) in the liver. We used a sample of monozygotic and dizygotic middle-aged male twins reared in the same family to account for familial and genetic confounding.

SUBJECTS AND METHODS

Participants

The Twins Heart Study is an investigation of psychological, behavioral, and biological risk factors for subclinical CVD using twins. This twin cohort was described previously (16). Briefly, this cohort includes 180 pairs of monozygotic and dizygotic male twins from the Vietnam Era Twin Registry, a registry of 7369 middle-aged male-male twin pairs, both of whom served in the US military during the Vietnam era (1964–1975). Twins selected for inclusion in the Twins Heart Study were born between 1946 and 1956, which represented >90% of the twins in the VET Registry. In addition, eligible twins were selected if they were without a history of symptomatic CVD, including a previous diagnosis of heart attack/myocardial infarction, coronary artery disease, angina, congestive heart failure or stroke, or previous coronary angioplasty or coronary bypass surgery according to self-reported data from a 1990 survey. For the Twins Heart Study, random samples of twins in 2 strata were selected from the Registry: one stratum included twins discordant for a lifetime history of major depression and the other included twins with no history of depression. Once selected, twin pairs were examined together at the General Clinical Research Center at Emory University between March 2002 and March 2006. The assessment included a comprehensive history and physical exam during which we obtained updated information about previous diagnoses of symptomatic CVD. Zygosity information, determined by DNA analysis, was available from all but 11 twin pairs. The zygosity of these 11 pairs was assessed by using questionnaires supplemented with blood group typing data abstracted from military records. We excluded 1 subject with no dietary data and 6 with implausible energy intake (≥6000 or <500 kcal/d). Therefore, our analyses were based on 353 twin participants (99 monozygotic and 74 dizygotic twin pairs and 3 monozygotic and 4 dizygotic unpaired twins). The study protocol was approved by the Institutional Review Board at Emory University, and informed consent was obtained from all participants.

Assessment of diet

We used the Willett self-administered semiquantitative food-frequency questionnaire (17) to collect habitual dietary intakes, including supplements over the previous 12 mo. The questionnaire classifies mean food intake according to 9 frequency categories ranging from “almost never or less than once per month” to “ ≥6 times/day.” Standardized portion sizes are used for each dietary item, including beverages and nutritional supplements. Questionnaires were scored by the Nutrition Questionnaire Service Center, Channing Laboratory, Harvard University, and nutrient intake data were derived by using the nutrient database of the US Department of Agriculture (17). The nutrient data used in the analysis included total energy intake and intakes of saturated fatty acids, monounsaturated fatty acids, trans-fatty acids, ALA, linoleic acid, arachidonic acid, γ-linolenic acid, and n−3 ≥20C-PUFAs (either from the diet or from fish-oil supplements: 20:5, 22:5, and 22:6). Daily food intake was calculated from food intake frequency and portion sizes.

Assessment of known CVD risk factors

We assessed smoking, education, and marital status using standardized questionnaires. Habitual physical activity, including occupational and leisure physical activity, was evaluated with the validated Baecke questionnaire (18). Waist and hip circumferences were measured and used to calculate the waist-to-hip ratio.

Height was measured by using a wall-mounted stadiometer measuring in 0.1-cm increments. Weight was measured by using a calibrated digital scale measuring in 0.1-kg increments. Height and weight were measured in fasting, postvoiding participants wearing light clothing, without shoes. Systolic and diastolic blood pressures were measured by using a mercury sphygmomanometer according to a standard protocol. Hypertension was defined as a systolic blood pressure ≥140 mm Hg and/or a diastolic blood pressure ≥90 mm Hg or the current use of antihypertensive medications. Diabetes was defined as a fasting plasma glucose concentration ≥6.93 mmol/L (19) or current treatment with insulin or oral hypoglycemic medications. Depressive symptoms were measured with the Beck Depression Inventory (BDI), which yielded a continuous score (20). Current use of statins, aspirin, and antihypertensive and antihyperglycemic medications was also recorded.

Biochemical analysis

Plasma samples were separated from 9-h overnight fasting blood samples and stored frozen at −80°C until analyzed. Laboratory personnel were blinded to subject identification and status, and twin pairs were assessed in the same analytic run. Fasting plasma concentrations of glucose, triglycerides, and total, LDL and HDL cholesterol were measured by using standardized methods. Fasting plasma concentrations of IL-6, total sIL-6R (free sIL-6R plus sIL-6R bound to IL-6), and TNF-α were measured by using commercial human enzyme-linked immunosorbent assay kits (Quantikine; R & D systems Inc, Minneapolis, MN) (21). The inter- and intraassay variability for all assays was <10%. CRP was measured with the Beckman Coulter High Sensitivity C-Reactive Protein (hsCRP) assay (Beckman Coulter Diagnostics, Brea, CA) on the Synchron LX-20 analyzer (Beckman Coulter Inc, Fullerton, CA). A plasma IL-6 concentration >10 pg/mL was defined as the cutoff for low-grade systemic inflammation, based on previous data (22).

Statistical analysis

Inflammatory biomarkers were log-transformed to improve normality. To estimate whether genetic factors influence dietary ALA intake and inflammatory biomarkers, intraclass correlation coefficients were calculated. We defined “within-pair absolute differences” as differences between a twin with a higher habitual dietary ALA intake and his twin brother with a lower intake.

The association between habitual dietary ALA intake and inflammatory biomarkers was assessed by fitting linear regression models adapted for twin studies at 2 levels (23): between-participants, ie, treating twins as separate individuals (without control for shared genes and environment) to represent the overall association and within-pair (with control for these shared factors) to reflect the unique effects of diet. Because dependent variables were natural log-transformed, the change in outcomes was expressed as the expected percentage difference in the nontransformed values for a 1-g increment in ALA intake in the overall association, a 1-g within-pair absolute difference in ALA intake between co-twins, or a 1-g between-pair difference in ALA intake between any 2 twin pairs by using the following formula:

where β is the regression coefficient and exp^β returns the exponential value of the exposure parameter (24). This outcome expression provides a convenient way to describe the healthy benefit associated with a 1-g dietary ALA intake. The latter is directly applicable to the dietary reference intake for ALA, which, in the United States, is 1.6 g/d for men older than 30 y (adequate intake).

At the between-subject level, we treated twins as individuals while accounting for clustering within a twin pair (23). Habitual dietary ALA intake was analyzed primarily as a continuous variable and secondarily as a categorical variable according to quintiles with intakes of 0.16 to 0.38, 0.39 to 0.53, 0.54 to 0.70, 0.71 to 0.91, and 0.92 to 2.31 g/d.

Next, we performed within-pair analyses to examine differences in inflammatory biomarkers between co-twins in each pair through partitioning the overall association into within- and between-pair associations. The within-pair effects were inherently matched for shared demographic, familial, and, in monozygotic pairs, genetic influences; additionally, environmental factors during the day of testing were controlled for because co-twins were examined at the same time. In these models, the within-pair β coefficient describes the individual twin variation from the twin pair mean of the exposure; this formulation is independent of twin ordering, and the results are identical to a model that fits the absolute difference between the co-twins (23). The percentage difference calculated from this parameter represents the percentage increment/decrement in biomarker concentrations per 1-g absolute difference in dietary ALA intake between twin brothers, comparing a twin with a 1-g higher ALA intake with his co-twin brother with a lower intake. The coefficient of between-pair effects, expressing variations in the twin pair mean of the exposure, reflects the influential magnitude of shared genes and common environment on the association between ALA intake and inflammation (23). Linear regression analyses included separated variance components for twin type to accommodate the different residual correlation in monozygotic and dizygotic pairs. Monozygotic and dizygotic twins were examined separately.

All of the covariates in the model were selected a priori. The “base model” included total energy intake and intakes of fatty acids inherently related to ALA metabolism in vivo, including linoleic acid (excluding trans- and trans-cis-linoleic acid) (continuous), γ-linolenic acid (continuous), arachidonic acid (continuous), n−3 ≥20C-PUFAs (including 20:5, 22:5, and 22:6 from fish-oil supplements) (continuous), and trans-, saturated, and monounsaturated fatty acids (excluding trans-monounsaturated fatty acids) (continuous). To this model we added sociodemographic factors [age (continuous), education (continuous), waist-to-hip ratio (continuous), physical activity (continuous), current smoking (yes or no), and marital status (yes or no)], clinical risk factors [depressive symptom score (continuous), plasma fasting glucose (continuous), systolic blood pressure (continuous), and LDL and HDL cholesterol (continuous)], and medication use [statins (yes or no), aspirin (yes or no), or antihypertensive (yes or no) or hypoglycemic (yes or no) agents]. To rule out potential multicollinearity, we transformed intakes of fatty acids other than ALA into corresponding z scores (difference between raw value and the sample mean divided by the SD) (25) and then diagnosed multicollinearity by using the condition index and variance decomposition proportions from an SAS macro for generalized estimating equation (25). The criteria used to judge multicollinearity are a condition index ≥30 and variance decomposition proportions ≥0.5 for ≥2 nonintercept predictors (26, 27). To rule out model overfitting, we fitted parsimonious models after backward elimination. However, our associations of interest from the parsimonious models and the full models were similar. In this study, we validated the stability and robustness of the within-pair association using 1000 bootstrap replicates. The association was robust if the within-pair association was statistically significant at α = 0.05 in ≥50% of full models fitted to the bootstrap replicates (28). All analyses were conducted by using SAS software version 9.1 (SAS Institute Inc, Cary, NC). Reported P values were 2-sided, and the type I error rate for significance was 0.05. Results are expressed as means ± SEMs or percentage geometric mean differences (95% CIs).

RESULTS

Sample characteristics

Participants with a higher habitual dietary intake of ALA had a higher total energy intake; had higher energy-adjusted intakes of linoleic acid, arachidonic acid, n−3 ≥20C-PUFAs, saturated and monounsaturated fatty acids; and were likely to use antihypertensive medications (Table 1). Only 3.7% of participants had an ALA intake ≥1.6 g/d—the US dietary reference intake (adequate intake) for men older than 30 y (29). The median ratio of dietary linoleic acid to ALA intake was 11 (range: 2.2–19.8) in monozygotic and 11 (range: 5.6–22.3) in dizygotic twins. A total of 63.7% participants in the pooled sample had a ratio >10 (range: 10.2–22.3) (29); to reach the ratio of 10, they would be expected to increase daily dietary ALA intake by a median of 0.12 g (range: 0.005–0.96) or by a median value of 0.22 g (range: 0.005–1.2) adjusted for a 2500-kcal caloric consumption. Plasma concentrations of sIL-6R were not related to IL-6 among monozygotic (P = 0.8) or dizygotic (P = 0.7) twins. The sample was 94% non-Hispanic whites, 3% African Americans, and 3% other racial-ethnic groups; this distribution reflected the racial distribution of the Vietnam Era Twin Registry from which it was sampled.

TABLE 1.

Characteristics of participants by quintiles of habitual dietary intake of α-linolenic acid (ALA)¹

	Quintile of ALA (g/d)
Variable	0.16–0.38	0.39–0.53	0.54–0.70	0.71–0.91	0.92–2.31	P for trend2
n	69	72	73	69	70
Nutritional factors
Total energy intake (kJ/d)	4105 ± 2383	5242 ± 234	6237 ± 234	7336 ± 238	9560 ± 238	<0.0001
Fatty acid intakes adjusted for total energy intake (g/d)
Linoleic acid4	4.73 ± 0.26	6.26 ± 0.24	6.62 ± 0.22	7.9 ± 0.24	11.35 ± 0.28	<0.0001
γ-Linolenic acid	0.01 ± 0.001	0.01 ± 0.001	0.01 ± 0.001	0.01 ± 0.001	0.01 ± 0.001	0.50
Arachidonic acid	7.4 ± 0.42	10.5 ± 0.38	11.0 ± 0.36	13.2 ± 0.39	17.6 ± 0.45	<0.0001
n−3 ≥20-carbon PUFAs
Including supplements	0.13 ± 0.03	0.13 ± 0.03	0.18 ± 0.03	0.22 ± 0.03	0.26 ± 0.03	<0.0001
Excluding supplements	0.12 ± 0.02	0.13 ± 0.02	0.16 ± 0.02	0.22 ± 0.02	0.21 ± 0.03	<0.0001
trans-Fatty acids	1.8 ± 0.1	2.0 ± 0.1	1.9 ± 0.1	1.8 ± 0.1	2.1 ± 0.1	0.09
Saturated fatty acids	21.8 ± 0.9	22.3 ± 0.8	22.2 ± 0.8	21.4 ± 0.8	23.5 ± 0.9	0.04
Monounsaturated fatty acids5	19.8 ± 0.8	21.7 ± 0.7	21.9 ± 0.7	22.1 ± 0.7	27.8 ± 0.9	<0.0001
Polyunsaturated fatty acids	5.99 ± 0.28	7.61 ± 0.25	8.13 ± 0.24	9.53 ± 0.26	13.5 ± 0.30	<0.0001
Sociodemographic factors
Age (y)	54.6 ± 0.3	54.7 ± 0.3	54.4 ± 0.3	54.5 ± 0.4	53.8 ± 0.3	0.67
Education (y)	14.2 ± 0.3	14.1 ± 0.2	14.6 ± 0.3	14.1 ± 0.2	14.3 ± 0.3	0.67
Married [n (%)]	48 (70)	55 (76)	64 (88)	56 (81)	56 (80)	0.18
Lifestyle factors
Waist-to-hip ratio	0.94 ± 0.01	0.95 ± 0.01	0.94 ± 0.01	0.94 ± 0.01	0.95 ± 0.01	0.20
BMI (kg/m2)	29.4 ± 0.5	29.3 ± 0.5	28.9 ± 0.5	28.8 ± 0.5	30.0 ± 0.6	0.18
Physical activity (unit)	7.1 ± 0.2	7.12 ± 0.2	7.7 ± 0.2	7.8 ± 0.2	7.3 ± 0.2	0.32
Current smokers [n (%)]	15 (22)	15 (21)	15 (21)	15 (22)	12 (17)	0.61
Clinical and biochemical features
Depression symptom score (BDI units)	5.1 ± 0.8	4.5 ± 0.7	5.1 ± 0.8	4.8 ± 0.8	4.9 ± 0.8	0.88
Plasma glucose (mmol/L)	5.39 ± 0.12	5.61 ± 0.12	5.56 ± 0.12	5.67 ± 0.12	5.45 ± 0.12	0.68
Systolic blood pressure (mm Hg)	130 ± 1.8	129 ± 1.8	129 ± 1.8	130 ± 1.8	129 ± 2	0.93
Diastolic blood pressure (mm Hg)	81 ± 1.2	80 ± 1.2	81 ± 1.2	81 ± 1.3	81 ± 1.3	0.61
Plasma lipids (mmol/L)
Total triglycerides	2.11 ± 0.12	1.91 ± 0.12	2.00 ± 0.12	2.00 ± 0.12	2.22 ± 0.14	0.11
Total cholesterol	4.84 ± 0.12	4.86 ± 0.12	4.86 ± 0.11	4.94 ± 0.12	4.73 ± 0.12	0.61
HDL cholesterol	1.03 ± 0.03	1.00 ± 0.03	0.99 ± 0.03	1.02 ± 0.03	0.96 ± 0.03	0.20
LDL cholesterol	3.10 ± 0.10	3.18 ± 0.10	3.15 ± 0.10	3.28 ± 0.10	3.13 ± 0.10	0.98
Use of statins [n (%)]	21 (30)	15 (21)	19 (26)	16 (23)	18 (26)	0.91
Use of aspirin [n (%)]	13 (19)	16 (22)	20 (27)	20 (29)	21 (30)	0.12
Use of antihypertensives [n (%)]	11 (16)	16 (22)	17 (23)	20 (29)	22 (31)	0.02
Use of antihyperglycemics [n (%)]	3 (4.4)	9 (12.5)	6 (8.2)	7 (10.1)	7 (10.1)	0.46

¹BDI, Beck Depression Inventory; PUFAs, polyunsaturated fatty acids.

²Test for trend across diet groups.

³Least-squares mean ± SEM estimated from linear mixed models accounting for clustering within a pair unless otherwise indicated (all such values).

⁴Excludes trans- and trans-cis-linoleic acid.

⁵Excludes trans-monounsaturated fatty acids.

Overall associations

Greater habitual dietary ALA intake was associated with lower plasma concentrations of sIL-6R, but not of IL-6, TNF-α, and hsCRP (Table 2). When habitual dietary ALA intake was treated as a continuous variable, higher habitual dietary ALA intake was significantly associated with lower plasma concentrations of sIL-6R after nutritional factors were controlled for (model A, Table 2). After full adjustment, the inverse association between habitual dietary ALA intake and sIL-6R remained strong (model B, Table 2). Similar results were observed when dietary ALA intake was treated as an ordinal variable (model B, Table 2). In the fully adjusted model, men in the highest quintile of dietary ALA intake had plasma sIL-6R concentrations 10.4% (95% CI: 1.5%, 19.3%) lower than those in the lowest quintile.

TABLE 2.

Overall association between habitual dietary α-linolenic acid (ALA) and plasma concentrations of inflammatory biomarkers¹

			Quintile of ALA (g/d)3
Outcome	Percentage difference2	P value	0.16–0.38	0.39–0.53	0.54–0.70	0.71–0.91	0.92–2.31	P value4
n			69	72	73	69	70
Model A5
IL-6 (pg/mL)	−20.7 (−49.9, 25.6)	0.32	1.84 (1.46, 2.31)	2.01 (1.66, 2.42)	1.77 (1.49, 2.09)	2.23 (1.85, 2.68)	1.82 (1.39, 2.40)	0.85
sIL-6R (ng/mL)	−11.0 (−17.7, −3.6)	0.005	28.5 (26.9, 30.5)	27.9 (26.3, 29.6)	28.2 (26.6, 29.7)	27.4 (25.8, 28.9)	25.5 (23.6, 27.4)	0.01
hsCRP (mg/L)	35.0 (−27.5, 151)	0.34	0.92 (0.66, 1.30)	1.08 (0.81, 1.45)	1.20 (0.92, 1.56)	1.52 (1.14, 2.01)	1.86 (1.23, 2.77)	0.12
TNF-α (pg/mL)	−11.7 (−38.3, 26.4)	0.50	1.28 (1.07, 1.55)	1.34 (1.14, 1.55)	1.39 (1.21, 1.60)	1.27 (1.09, 1.47)	1.28 (1.03, 1.60)	0.91
Model B6
IL-6 (pg/mL)	−25.3 (−52.9, 18.3)	0.21	2.36 (1.80, 3.12)	2.53 (2.01, 3.22)	2.25 (1.80, 2.83)	2.69 (2.15, 3.39)	2.08 (1.53, 2.79)	0.35
sIL-6R (ng/mL)	−11.1 (−17.8, −3.9)	0.003	28.5 (26.2, 30.7)	27.7 (25.7, 29.8)	28.2 (26.2, 30.2)	26.8 (25.1, 28.9)	25.3 (23.3, 27.5)	0.011
hsCRP (mg/L)	59.5 (−13.6, 195)	0.14	1.11 (0.74, 1.67)	1.20 (0.84, 1.72)	1.42 (1.00, 2.02)	1.57 (1.10, 2.23)	1.86 (1.18, 2.91)	0.23
TNF-α (pg/mL)	−19.3 (−44.2, 16.6)	0.25	1.35 (1.08, 1.69)	1.38 (1.13, 1.68)	1.39 (1.15, 1.68)	1.28 (1.06, 1.56)	1.16 (0.91, 1.50)	0.37

¹IL-6, interleukin-6; TNF-α, tumor necrosis factor-α; hsCRP, high-sensitivity C-reactive protein; sIL-6R, soluble receptor of IL-6.

²Values are percentage geometric mean differences (95% CIs); n = 353 men. The percentage was calculated from the β coefficient for a 1-g increment in ALA intake. The linear mixed model was used for the analyses and accounted for clustering within the twin pair while allowing for different correlations in monozygotic and dizygotic pairs. A negative value indicates that an individual with a 1-g higher ALA intake was likely to have lower concentrations of biomarkers than an individual with a 1-g lower intake. Conversely, a positive value indicates that an individual with a 1-g higher ALA intake was likely to have higher concentrations of biomarkers than an individual with a 1-g lower intake.

³Values are geometric means (95% CIs).

⁴Test for trend across diet groups accounting for clustering within a pair by twin types. An ordinal ALA variable was generated with mean ALA intake as the rank value.

⁵Adjusted for total energy intake, linoleic acid (excluding trans- and trans-cis-linoleic acid), γ-linolenic acid, arachidonic acid, n−3 ≥20-carbon polyunsaturated fatty acids (including 20:5, 22:5, and 22:6 from fish-oil supplements), and trans-, saturated, and monounsaturated fatty acids (excluding trans-monounsaturated fatty acids).

⁶Additionally controlled for age, education, waist-to-hip ratio, physical activity, current smoking, marital status, depressive symptom score, plasma glucose, systolic blood pressure, LDL and HDL cholesterol, and use of statins, aspirin, and antihypertensive and hypoglycemic medications.

Within-pair association

The within-pair association between dietary ALA intake and plasma concentrations of all biomarkers was not statistically significantly different by zygosity (all P > 0.25), so we pooled samples by zygosity. The within-pair inverse association between dietary ALA intake and plasma concentrations of sIL-6R was statistically significant before and after adjustment for known CVD risk factors, whereas there was no association between dietary ALA intake and plasma concentrations of IL-6, TNF-α, or hsCRP (Table 3). This association was further confirmed among monozygotic pairs, for whom genetic factors were completely controlled for. Overall, the results remained similar after exclusion of participants with previous coronary heart disease and/or IL-6 concentrations >10 pg/mL (data not shown).

TABLE 3.

Within- and between-pair percentage differences in plasma concentrations of inflammatory biomarkers per 1-g increment in habitual dietary α-linolenic acid intake in male twins¹

	Within-pair difference (95% CI)2							Between-pair difference (95% CI)3
	MZ + DZ (n = 353)		MZ (n = 201)		DZ (n = 152 )		P for interaction with zygosity	MZ + DZ (n = 353)
Outcome	Percentage	P value	Percentage	P value	Percentage	P value		Percentage	P value
Model A4
IL-6 (pg/mL)	−22.9 (−51.6, 22.9)	0.27	−20.8 (−55.6, 41.6)	0.43	−25.2 (−66.8, 68.2)	0.48	0.58	−1.0 (−40.5, 64.6)	0.97
sIL-6R (ng/mL)	−11.3 (−18.0, −4.0)	0.003	−11.4 (−18.6, −3.4)	0.006	−15.8 (−31.9, 4.2)	0.11	0.49	−17.1 (−27.9, −4.7)	0.009
hsCRP (mg/L)	29 (−30, 140)	0.42	25 (−39, 158)	0.54	16 (−67, 311)	0.82	0.61	64 (−23, 251)	0.20
TNF-α (pg/mL)	−12.2 (−38.4, 25.2)	0.47	−24.8 (−55.2, 26.1)	0.28	4.5 (−36.5, 71.8)	0.86	0.90	−6.0 (−37.2, 40.0)	0.76
Model B5
IL-6 (pg/mL)	−30.6 (−56.8, 11.3)	0.13	−9.7 (−49.8, 62.4)	0.73	−56.3 (−80.9, 0.3)	0.051	0.27	−14.4 (−48.6, 42.5)	0.55
sIL-6R (ng/mL)	−11.0 (−17.7, −3.8)	0.003	−12.4 (−19.3, −4.8)	0.002	−15.8 (−32.6, 5.2)	0.13	0.39	−15.9 (−26.8, −3.3)	0.02
hsCRP (mg/L)	51 (−19, 180)	0.19	61 (−21, 229)	0.18	−42 (−85, 129)	0.43	0.33	112 (0.2, 347)	0.050
TNF-α (pg/mL)	−19.0 (−44.3, 17.9)	0.27	−30.4 (−58.7, 17.1)	0.17	−12.5 (−50.3, 53.9)	0.64	0.98	−20.4 (−47.5, 21)	0.28

¹MZ, monozygotic; DZ, dizygotic; IL-6, interleukin-6; TNF-α, tumor necrosis factor-α; hsCRP, high-sensitivity C-reactive protein; sIL-6R, soluble receptor of IL-6.

²Values are within-pair percentage geometric mean differences (95% CIs) per 1-g within-pair difference in α-linolenic acid intake; n = 99 MZ and 74 DZ twin pairs and 3 MZ and 4 DZ unpaired twins. The within-pair difference was calculated from the β coefficient for within-pair effects and is expressed per 1-g difference in the α-linolenic acid intake between co-twins within a pair.

³The between-pair difference was calculated from the β coefficient for between-pair effects and is expressed per 1-g difference in the α-linolenic acid intake between any 2 pairs of twins. The linear mixed model was used for the analyses and accounted for clustering within the twin pair allowing for different correlations in MZ and DZ pairs.

⁴Adjusted for total energy intake, linoleic acid (excluding trans- and trans-cis-linoleic acid), γ-linolenic acid, arachidonic acid, n−3 ≥20-carbon polyunsaturated fatty acids (including 20:5, 22:5, and 22:6 from fish-oil supplements), and trans-, saturated, and monounsaturated fatty acids (excluding trans-monounsaturated fatty acids).

⁵Additionally controlled for age, education, waist-to-hip ratio, physical activity, current smoking, marital status, depressive symptom score, plasma glucose, systolic blood pressure, LDL and HDL cholesterol, and use of statins, aspirin, and antihypertensive and hypoglycemic medications.

In secondary analyses, we found a significant association between the ratio of linolenic to ALA (linolenic acid:ALA) and sIL-6R, but not to other biomarkers. In the sample pooled by zygosity, a twin with a 1-unit higher ratio was likely to have a 1% (95% CI: 0.4, 1.6; P = 0.001) higher sIL-6R concentration than his twin brother after control for shared genes, common environment, medication use, total energy intake, and other nutritional, sociodemographic, and clinical coronary heart disease risk factors. The association was similar by zygosity (P = 0.93 for interaction between the linolenic acid:ALA ratio and zygosity on sIL-6R).

Intake of n−3 ≥20C-PUFAs, including eicosapentaenoic acid, docosapentaenoic acid, and docosahexaenoic acid, may inhibit ALA metabolism through a feedback mechanism; therefore, it was adjusted for in our multivariable analyses. However, to better understand the role of these nutrients, we also conducted a within-pair analysis of the association of n−3 ≥20C-PUFA intake with inflammatory biomarkers. In the sample pooled by zygosity, a 1-g higher habitual dietary intake of n−3 ≥20C-PUFAs was associated with a 10.7% higher sIL-6R concentration (P = 0.02) after control for shared genes and common environment (by design), medication use, and other nutritional, sociodemographic, and clinical coronary heart disease risk factors. The association was similar by zygosity (P for interaction term = 0.26).

Between-pair association

The between-pair association between dietary ALA intake and plasma concentrations of all biomarkers was not significantly different by zygosity (all P > 0.1), so we pooled samples by zygosity. Between-pair associations of ALA intake with sIL-6R and hsCRP were statistically significant in the full model (Table 3, model B).

The frequency of statistically significant associations of within- and between-pair effects of ALA with sIL-6R in the full model was >50% among 1000 bootstrap replicates (Table 4). The estimate of the magnitude and direction of the within- and between-pair association for sIL-6R from the bootstrap samples was similar to that from the original data set.

TABLE 4.

Bootstrap validation for within- and between-pair associations between habitual α-linolenic acid intake and inflammatory biomarkers in the full model by using 1000 bootstrap replicates¹

	Frequency for P ≤ 0.05	β Estimates (95% CI)2	Difference (95% CI)3
IL-6 (pg/mL)
Within-pair	33.2	−0.336 (−0.817, 0.145)	−28.5 (−55.8, 15.6)
Between-pair	3.9	−0.050 (−0.550, 0.459)	−4.7 (−42.3, 48.0)
sIL-6R (ng/mL)
Within-pair	78.3	−0.132 (−0.235, −0.03)	−12.3 (−20.9, −2.9)
Between-pair	52.0	−0.164 (−0.310, 0.037)	−15.1 (−26.7, 3.8)
hsCRP (mg/L)
Within-pair	25.6	0.284 (−0.551, 1.106)	33 (−42.3, 202)
Between-pair	44.1	0.742 (−0.107, 1.555)	110 (−10, 373)
TNF-α (pg/mL)
Within-pair	16.1	−0.174 (−0.521, 0.195)	−16.0 (−40.6, 21.5)
Between-pair	10.5	−0.201 (−0.632, 0.190)	−18.2 (−46.8, 20.9)

¹IL-6, interleukin-6; TNF-α, tumor necrosis factor-α; hsCRP, high-sensitivity C-reactive protein; sIL-6R, soluble receptor of IL-6. For each bootstrap replicate, the full model was adjusted for nutritional [total energy intake, linoleic acid (excluding trans- and trans-cis-linoleic acid), γ-linolenic acid, arachidonic acid, n−3 ≥20-carbon polyunsaturated fatty acids (including 20:5, 22:5, and 22:6 from fish-oil supplements), and trans-, saturated, and monounsaturated fatty acids (excluding trans-monounsaturated fatty acids)]; sociodemographic (age, education, and marital status), lifestyle (waist-to-hip ratio, physical activity, and current smoking), and clinical (depressive symptom score, plasma glucose, systolic blood pressure, and LDL and HDL cholesterol) risk factors; and medication use (use of statins, aspirin, or antihypertensive and hypoglycemic medications).

²Expected changes in natural log-transformed concentrations of biomarkers for a 1-g difference in α-linolenic acid either between co-twins or between the mean of any 2 twin pairs. The linear mixed model was used for the analyses and accounted for clustering within the twin pair.

³The within-pair difference (%) was calculated from the β coefficient for within-pair effects and was expressed per 1-g difference in the α-linolenic acid intake between co-twins within a pair. The between-pair difference (%) was calculated from the β coefficient for between-pair effects and was expressed per 1-g difference in the α-linolenic acid intake between any 2 pairs of twins.

DISCUSSION

We found a strong inverse within-pair association of habitual dietary ALA intake with plasma sIL-6R concentrations, independent of a wide range of known CVD risk factors in this middle-aged male twin cohort: a 1-g higher habitual dietary ALA intake was significantly associated with an 11.0% (95% CI: 3.8%, 17.7%) lower plasma concentration of sIL-6R. The association persisted when twins within a monozygotic pair were compared. These findings indicate that the association between habitual dietary ALA and sIL-6R is independent of shared genes and common environment. Adjustment for confounding by shared genes and common environment, through within-pair analyses, did not change the direction of association between ALA intake and inflammatory biomarkers. After this adjustment, the association of ALA intake with IL-6 and hsCRP was strengthened by 25% and 12%, respectively, but still remained statistically insignificant, whereas the association of ALA intake with sIL-6R and TNF-α was slightly attenuated by 1% and 1.7%, respectively. We also found a strong between-pair association between habitual dietary ALA intake and plasma sIL-6R concentrations with a magnitude and direction similar to the within-pair association. Because the within-pair effects of ALA on sIL-6R did not differ by zygosity and shared genes did not play a role in the association between ALA intake and sIL-6R, the significant between-pair effect may be attributable to environmental factors shared between co-twins but not genes (23). The within- and between-pair associations between dietary ALA and sIL-6R found in the original data set were validated with 1000 bootstrap replicates.

The soluble IL-6 receptor (sIL-6R) enhances IL-6–mediated inflammation (4). Although the synergistic role of sIL-6R and IL-6 is clear at the cellular level, the association of sIL-6R with CVD and its risk factors in humans is much less known. In a recent study, higher serum sIL-6R concentrations were associated with more severe sleep-related breathing disorders—a CVD risk factor (30). IL-6R concentrations in humans’ atherosclerotic plaques are positively associated with CVD risk factors, including smoking and diabetes (31). A paradoxically adverse relation between blood cholesterol concentrations and CVD mortality in hemodialysis patients has been attributed to less sIL-6R released from activated peripheral blood mononuclear cells (32). These previous findings primarily link sIL-6R to CVD risk in patient populations. Our results suggest that sIL-6R should be examined more thoroughly in future research as a potential novel risk factor for CVD in predominantly healthy individuals.

The association between habitual dietary ALA intake and plasma sIL-6R concentrations has rarely been examined. One 4-d randomized trial among healthy adults found that a diet in which ALA provided 5% of total energy intake lowered plasma sIL-6R concentrations compared with a control diet in which ALA provided 0.5% of total energy intake (33). Our study showed for the first time that this finding applies also to habitual ALA dietary intake at much lower concentrations of dietary ALA, which in our study ranged from 0.16 to 2.3 g/d (from 0.14% to 0.93% of total energy intake).

It is surprising that the intake of n−3 ≥20C-PUFAs was positively associated with plasma concentrations of sIL-6R, whereas sIL-6R was inversely associated. Ferrucci et al reported a similar finding (34). If true, the biological significance of this association will need further evaluation.

Although we found a robust association between dietary ALA and sIL-6R, we did not find a statistically significant association between habitual dietary ALA intake and plasma concentrations of IL-6, TNF-α, and CRP. The association of dietary ALA intake with circulating concentrations of IL-6, TNF-α, and CRP has been examined in previous studies including ex vivo experiments (10, 35, 36), observational reports (5, 6), and randomized trials (7–9, 37, 38), and the results are inconsistent. The randomized trial with the largest sample size and the longest intervention duration (2 y) reported that an intake of 5.9 g ALA/d (2.3% of total energy intake) did not significantly affect IL-6 concentrations compared with 1 g ALA/d (0.4% of total energy intake) (7), which were consistent with our study and another population study in the United States (5). Our findings of a lack of association between blood concentrations of IL-6, TNF-α, CRP, and ALA intake were also in agreement with several other studies (6, 10).

Inflammatory responses involve various types of cells that can be activated through multiple pathways via the action of IL-6 on cellular receptors (39). Because endothelial and smooth muscle cells rely on sIL-6R to respond to IL-6 (40), our findings suggest that sIL-6R may be a more sensitive and novel indicator of inflammatory processes in tissues most affected by atherosclerosis (41). For example, sIL-6R, but not IL-6, stimulates endothelial cells to synthesize cellular adhesion molecules and to bind neutrophils, which, in turn, promote endothelial cells to recruit more leukocytes by shedding considerable amounts of sIL-6R (41).

The possible underlying biochemical mechanism through which habitual dietary ALA lowers sIL-6R is unclear. In vivo, sIL-6R produced by shedding (proteolytic cleavage), but not from differential mRNA splicing, is regulatory (4). sIL-6R shedding is triggered by elevated intracellular calcium concentrations, (42) which, in turn, are probably regulated by arachidonic acid (43, 44). ALA decreases arachidonic acid concentrations by competing with linoleic acid for common enzymes (1). Therefore, dietary ALA might putatively reduce sIL-6R shedding by decreasing arachidonic acid concentrations.

Our study had a number of limitations. The Willett food-frequency questionnaire may not be optimal for estimating absolute intakes because it can underestimate nutrient intakes. However, its use was appropriate in our investigation, in which we assessed associations with diet after energy intake adjustment (45). Participants with higher dietary intakes of ALA were more likely to have higher intakes of trans-fatty acids, linoleic acid, and n−3 long-chain, saturated, and monounsaturated fatty acids. To minimize confounding, we controlled for these and other factors, making it less likely that they confound the association. The sample was restricted to middle-aged men, and our results may not be generalizable to females. The men in our sample had a relatively high mean BMI; thus, our results may not be generalizable to populations with lower mean BMI. However, the BMI distribution in our sample is similar to that of men from the general US population (46). We did not measure plasma concentrations of sgp 130, a natural inhibitor of IL-6/sIL-6R/gp 130 trans-signaling. It is, therefore, unknown whether the association between ALA intake and sIL-6R was independent of sgp 130. Finally, we did not measure plasma concentrations of fatty acids. Thus, we could not test the mediation of plasma fatty acids in the association of dietary ALA intake with inflammation. Nevertheless, our study was strengthened by the use of a twin design to control for unmeasured and unknown confounding variables, such as genetic factors; socioeconomic, behavioral, and lifestyle characteristics; and maternal/prenatal factors acquired by twins reared within the same family.

In conclusion, we found that habitual dietary ALA intake was inversely associated with plasma sIL-6R concentrations independent of shared genetic and common environmental factors. Our findings support the hypothesis that sIL-6R is a novel biomarker for the cellular inflammatory response to dietary ALA. Our results are relevant from a clinical and public health standpoint, because a 1.0-g increase in daily dietary ALA intake is easily achievable. One tablespoon (15 mL) of canola oil, nonhydrogenated soybean oil, or half a teaspoon (2.5 mL) of linseed (flaxseed) oil provides ≈1 g ALA. Our findings support the potential importance of increasing dietary ALA in the habitual diet to prevent CVD.