Associations between plasma polyunsaturated fatty acids, plasma stearoyl-CoA desaturase indices and body fat

Authors

  • Kathrine J. Vinknes,

    Corresponding author
    • Department of Nutrition, Institute of Basic Medical Science, Faculty of Medicine, University of Oslo, Oslo, Norway
    Search for more papers by this author
  • Amany K. Elshorbagy,

    1. Department of Pharmacology, University of Oxford, Oxford, UK
    2. Department of Physiology, Faculty of Medicine, University of Alexandria, Alexandria, Egypt
    Search for more papers by this author
  • Christian A. Drevon,

    1. Department of Nutrition, Institute of Basic Medical Science, Faculty of Medicine, University of Oslo, Oslo, Norway
    Search for more papers by this author
  • Eha Nurk,

    1. Department of Nutrition, Institute of Basic Medical Science, Faculty of Medicine, University of Oslo, Oslo, Norway
    2. Department of Surveillance and Evaluation, National Institute for Health Development, Tallinn, Estonia
    Search for more papers by this author
  • Grethe S. Tell,

    1. Department of Public Health and Primary Health Care, University of Bergen, Bergen, Norway
    Search for more papers by this author
  • Ottar Nygård,

    1. Section for Cardiology, Institute of Medicine, University of Bergen, Bergen, Norway
    2. Department of Heart Disease, Haukeland University Hospital, Bergen, Norway
    Search for more papers by this author
  • Stein E. Vollset,

    1. Department of Public Health and Primary Health Care, University of Bergen, Bergen, Norway
    2. Division of Epidemiology, Norwegian Institute of Public Health, Oslo, Norway
    Search for more papers by this author
  • Helga Refsum

    1. Department of Nutrition, Institute of Basic Medical Science, Faculty of Medicine, University of Oslo, Oslo, Norway
    2. Department of Pharmacology, University of Oxford, Oxford, UK
    Search for more papers by this author

  • Disclosure: CAD is a cofounder, consultant and board member of the analytical company Vitas AS (www.vitas.no). CAD owns 8% of the shares, and other family members own 16% of the shares in Vitas. KJE, AKE, EN, GST, ON, SE and HR had no conflicts of interest.

  • Funding agencies: This study has received support from the Advanced Research Programme of Norway, The Research Council of Norway, Norwegian Rheumatism Association, The Johan Throne Holst Foundation for Nutrition Research, and University of Oslo, Norway. None of the funding bodies were involved in the study design, collection, analyses, interpretation of the data, or preparation of the manuscript.

  • Author contributions: KJV, AKE, and HR: designed and conducted the research; CAD, GST, SEV, and HR: data collection; KJV, AKE, and HR: analyzed data and performed statistical analyses; KJV and HR: wrote the article; CAD organized development of the FFQ; AKE, EN, CAD, GST, ON and SE: critically revised manuscript for important intellectual content; and KJV and HR: had primary responsibility for final content.

Correspondence: Kathrine J. Vinknes (kathrine.vinknes@medisin.uio.no)

Abstract

Objective

Stearoyl-CoA desaturase (SCD)-1 deficient mice are resistant to obesity and plasma SCD indices are related to obesity in humans. Both n-3 and n-6 polyunsaturated fatty acids (PUFA) regulate expression of the SCD enzymes. Whether higher plasma PUFA were associated with lower SCD indices in humans was examined.

Design and Methods

Population-based study of 2,021 elderly subjects from the Hordaland Health Study. Using multivariate linear regression, the cross-sectional associations among plasma PUFA, estimated SCD indices (from fatty acid profiles in plasma total lipids), and fat mass measured by dual-energy X-ray absorptiometry were explored. Two plasma SCD indices were used: SCD-16 (16:1n-7/16:0) and SCD-18 (18:1n-9/18:0).

Results

Plasma total, n-6 and n-3 PUFA were inversely associated with both SCD indices (P < 0.001 for all). Among the individual PUFA, 18:2n-6 showed the strongest association with SCD-16 (partial r = −0.59, P < 0.001) followed by 20:5n-3 (partial r = −0.13; P < 0.001). Plasma total, n-6 and n-3 PUFA were inversely associated with body fat (P < 0.001 for all); the associations were markedly attenuated following adjustment for SCD-16.

Conclusions

The epidemiological data are in line with animal studies and suggest that PUFA may decrease SCD1 activity in humans, with possible reduction in body fat.

Introduction

Stearoyl-CoA desaturase (SCD) is a key enzyme regulating whole body lipid composition, by introducing a cis-double bond at the delta-9 position in the backbone of 12–19 carbon saturated fatty acids (SFA), resulting in the production of monounsaturated fatty acids (MUFA) [1]. High SCD1 activity has been shown to play a role in the development of obesity and related conditions in animal models [2]. Global SCD1-knockout mice are protected from high-fat diet obesity and insulin resistance [2], although liver-specific and adipose tissue-specific knockout mice are not [3]. In human studies, indirect plasma indices of SCD activity are commonly used by estimating the ratio of oleic acid to stearic acid (18:1n-9/18:0) or palmitoleic acid to palmitic acid (16:1n-7/16:0) [4], and positive correlations between plasma SCD indices and adiposity have been reported [5, 6]. Because the 16:1n-7/16:0 ratio is less susceptible to dietary influence, than the 18:1n-9/18:0 ratio, the 16:1n-7/16:0 ratio has been suggested as a preferred marker of SCD1 activity [4, 7]. An ∼8 kg increase in fat mass from low to high plasma SCD index (16:1n-7/16:0) was recently reported in the Hordaland Health Study (HUSK) [8].

The type and quantity of fat consumed regulates hepatic lipid composition and gene expression [9]. Polyunsaturated fatty acids (PUFA) regulate expression of the enzymes involved in synthesis and oxidation of SFA and MUFA through key transcription factors [10]. The repression of SCD transcription by both n-3 and n-6 PUFA depends on sterol regulatory element binding proteins (SREBP-1c) [10, 11]. n-3 PUFA are reported to be more potent ligands to this transcription factor than n-6 PUFA [12]. In mice diets containing either n-3 PUFA-rich fish oil or n-6 PUFA-rich soy bean oil (linoleic acid; 18:2n-6) repressed SCD1 expression [13].

Intake of PUFA has been shown to improve body composition in humans, and especially long chain n-3 PUFA has received much attention in this context [reviewed in Ref. [14, 15]]. Small-cross-sectional studies have shown inverse relationships between blood long chain n-3 PUFA with BMI and total body adipose tissue [16, 17]. Conversely, it has been suggested that n-6 PUFA promote adipogenesis in vitro and adipose tissue development in animals [18, 19], although there is no evidence for a causal link between n-6 PUFA intake and human obesity [20]. Thus the relation of n-3 and n-6 PUFA status with fat mass in humans needs to be evaluated in larger studies.

The mechanism by which PUFA may influence body composition is unclear, but experimental studies have reported n-3 and n-6 PUFA to be repressors of SCD1 transcription [9, 21]. In our recent report of dietary determinants of plasma SCD indices in over 2,000 subjects from HUSK, higher n-3 and n-6 PUFA intakes were indeed associated with lower SCD indices [8]. In the present study we extend these findings by exploring the associations of individual and total n-3 and n-6 PUFA concentrations in plasma with SCD indices and fat mass in the same population.

Subjects and Methods

Study population

HUSK was conducted from 1997 to 1999 as a collaboration between the University of Bergen, the National Health Screening Service, local health services in the Bergen area and the University of Oslo. Details of the study are described elsewhere [22]. The present cross-sectional study is based on data on the fatty acid profile available from 924 men and 1,097 women in the HUSK born in 1925-1927.

All participants gave their written, informed consent. The study protocol was approved by the Regional Committee for Medical Research Ethics of Western Norway, based on directives in the Helsinki declaration.

Body composition measurements

Fat mass was assessed by dual-energy X-ray absorptiometry (DXA), which is based on the differential attenuation of photons by different types of body tissues. Transmission of X-rays at two energy levels allows the derivation of total body bone mineral mass, lean mass, and fat mass [23]. Measurements were performed on a stationary fan-beam densitometer using EXPERT-XL software (version 1.72–1.9; Lunar Corporation, Madison, WI). The coefficient of variation for fat mass was 1.9%.

Lifestyle and dietary data

Self-administered questionnaires provided information on diet (FFQ) and lifestyle. Nutrient intakes were calculated using a food database and software system developed at the Institute for Nutrition Research, University of Oslo (KOSTBEREGNINGSSYSTEM, version 3.2: University of Oslo, Norway) [24]. The validity of the reported dietary habits used in the study has been evaluated against 14-day weighed records and biomarkers [24]. Physical activity included two variables referring to heavy physical activity or light physical activity in the past year, where each variable comprised four categories: (1) none, (2) <1 h week−1, (3) 1-2 h week−1, and (4) ≥3 h week−1. The category numbers were replaced by a summary score estimated for each subject as previously described [8]. Current smoking and coffee consumption were used as continuous variables comprising the number of cigarettes smoked per day or the consumption of coffee in gram/day.

Biochemical measurements

Nonfasting blood samples were collected for measurement of serum lipids and for preparation of EDTA plasma [25]. Fatty acid profiles in plasma total lipids were analyzed by gas liquid chromatography with flame ionization detection at AS Vitas, Oslo Innovation Center, Norway (www.vitas.no) described previously [8]. Fatty acid content was calculated based on the area % of peaks and response factors relative to stearic acid (18:0). A gas liquid chromatogram of fatty acid methyl esters (FAME) in human plasma expressing PUFA from 18:2n-6 to 22:6n-3 is presented in Supporting Information Figure S1. Plasma SCD indices were estimated as product/precursor ratios of fatty acids in plasma [7]. SCD-16 index was calculated as 16:1n-7/16:0 and SCD-18 index as 18:1n-9/18:0.

Statistical analyses

Associations of plasma PUFA with plasma SCD indices and fat mass were estimated by linear regression analyses. We focused in the regression models on four selected PUFA that were quantitatively important, and with partial r > ± 0.15 in a stepwise regression analysis including all PUFA. None of the other PUFA had higher partial r. The selected PUFA included linolenic acid (18:3n-3), eicosapentanoic acid (20:5n-3), linoleic acid (18:2n-6), and arachidonic acid (20:4n-6). Initially, interactions were investigated to determine whether sex or n-6 and n-3 PUFA status modified the associations of plasma PUFA with plasma SCD indices. We included a product term of the plasma SCD indices and sex. There was no significant interaction for either index, and in an analysis stratified by sex, the patterns were nearly similar in both sexes. Thus, analyses were performed for the combined group including men and women. Furthermore, there was no interaction between n-6 and n-3 PUFA examined as continuous variables or quartiles. We used multivariate linear regression models to examine the association of plasma PUFA with plasma SCD indices. We then investigated whether plasma PUFA were associated with fat mass, and if this relation was modified by adjusting for plasma SCD indices.

Plasma SCD-16 and SCD-18 indices and fat mass were the outcome variables in regression models that included either total PUFA, n-6 and n-3 PUFA, or all the individual PUFA, together with potential confounders. We mutually adjusted for n-3 and n-6 PUFA to assess their independent effects. We conducted three models: The covariate included in Model 1 was sex. Model 2 included sex, triglycerides (TG) and cholesterol, and Model 3 included variables from Model 2, in addition to intake of energy, alcohol, and coffee, physical activity score, smoking, and time since last meal. The covariates included in Models 2 and 3 were selected based on their significant correlations both with the plasma SCD indices and plasma concentration of PUFA. Because of the narrow age range (2 years), age was not included. Diabetes or history of cardiovascular disease did not change the models appreciably and were excluded from the final analyses. Intakes of fat, carbohydrate and sugar were not included in the models unless otherwise stated.

We used Gaussian generalized additive regression models (R foundation for Statistical Computing, version 2.13.0, Vienna, Austria) to create dose-response curves of the differences in plasma SCD-16 and SCD-18 indices by plasma PUFA (total PUFA, n-6 PUFA, 18:2n-6, 20:4n-6, n-3 PUFA, 18:3n-3 and 20:5n-3), and the difference in fat mass by plasma total PUFA, n-3 PUFA and n-6 PUFA. At approximately mean exposure of the independent variable, the model generates a reference value of zero for the dependent variable. Models with different covariates are specified in the figure legends. Corresponding P values were obtained from multiple linear regression analyses.

Except for generalized additive models, all statistical analyses were performed with PASW Statistics for WINDOWS (18.0; SPSS, Chicago, IL). Tests of significance were two-tailed and P < 0.05 was considered significant.

Results

Characteristics of the study population

Characteristics of the study population have been reported previously [8] (Supporting Information Table S1). The participants were aged 71-74 years, and 45.7% were males. Average BMI was 26.0 kg m−2 with no significant difference between men and women, whereas body fat percent was 27.0% for men and 40.5% for women. Diabetes was present in 6.5% of the study population, and a total of 33.6% of the participants had reported history of cardiovascular disease. The study population had a high daily fish and fish oil intake [26]. In general, the distribution of energy between the macronutrients (% of total energy) were in line with the Nordic Nutrition recommendations [27], except that energy from SFA was slightly higher than the recommended <10% of the total energy intake.

The fatty acid profile

The fatty acid profile in total plasma lipids expressed as a percentage of total fatty acids [g/100 g fatty acid methyl esters (FAME)] is presented in Table 1. The 18:2n-6 constituted the highest proportion of total fatty acids followed by 16:0, 18:1n-9, and 18:0. Among the n-3 PUFA, 22:6n-3, 20:5n-3, and 18:3n-3 constituted the highest proportions of total fatty acids. Among the n-6 PUFA, 18:2n-6 and 20:4n-6 were the major components. The relative content of 16:0, eicosenoic acid (1n-9), 18:2n-6, eicosadienoic acid (20:2n-6), 18:3n-3, 20:5n-3, 22:6n-3, 18:2n-6, and eicosadienoic acid (20:2n-6) were significantly higher in men, whereas women had significantly higher proportions of pentadecanoid acid (15:0), 18:0, arachidic acid (20:0), behenic acid (22:0), tricosyclic acid (23:0), 16:1n-7, 18:1n-9, gamma-linoleic acid (18:3n-6), dihomo-y-linoleic acid (20:3n-6), and 20:4n-6.

Table 1. The fatty acid profile in the Hordaland Health Study subjects (% of total fatty acids)
 Total population (n = 2021)Men (n = 924)Women (n = 1097)P valuea
  1. MUFA, monounsaturated fatty acids; PUFA, polyunsaturated fatty acids; SCD, stearoyl-coenzyme A desaturase; SFA, saturated fatty acids; vcl, very-long-chain-n-3 (includes eicosapentaenoic, docosapentaenoic and docosahexaenoic acid).

  2. a

    Independent Sample t-test, P < 0.05.

  3. b

    Values are means (SD).

  4. c

    PUFA selected for analysis in this study.

SFA32.77 (1.86)b32.81 (1.92)32.74 (1.81)0.42
C14:0 (myristic acid)1.23 (0.38)1.21 (0.40)1.24 (0.37)0.070
C15:0 (pentadecanoid acid)0.23 (0.05)0.23 (0.05)0.24 (0.05)<0.001
C16:0 (palmitic acid)21.76 (1.62)21.94 (1.66)21.62 (1.56)<0.001
C18:0 (stearic acid)7.63 (0.63)7.57 (0.63)7.69 (0.62)<0.001
C20:0 (arachidic acid)0.29 (0.05)0.28 (0.05)0.30 (0.05)<0.001
C22:0 (behenic acid)0.76 (0.24)0.74 (0.24)0.78 (0.24)0.001
C23:0 (tricosanoic acid)0.31 (0.06)0.29 (0.06)0.32 (0.06)<0.001
C24:0 (Lignoceric acid)0.55 (0.11)0.54 (0.12)0.55 (0.11)0.35
MUFA22.46 (2.65)22.11 (2.72)22.75 (2.56)<0.001
C16:1n-7 (palmitoleic acid)2.09 (0.68)1.90 (0.66)2.24 (0.65)<0.001
C18:1n-9 (oleic acid)16.67 (2.16)16.51 (2.21)16.81 (2.11)0.002
C18:1t6-t11 (elaidic Acid=t9)1.15 (0.42)1.13 (0.44)1.15 (0.41)0.29
C18:1c11n-7 (vaccenic acid)1.37 (0.20)1.37 (0.19)1.36 (0.20)0.34
C20:1n-9 (eicosenoic acid)0.14 (0.09)0.16 (0.10)0.13 (0.07)<0.001
C24:1n-9 (Nervonic acid)1.05 (0.28)1.04 (0.29)1.05 (0.28)0.49
PUFA38.4 (3.96)38.86 (4.16)38.01 (3.75)<0.001
n-6 PUFA31.88 (4.16)32.16 (4.57)31.65 (3.77)0.006
C18:2n-6 (linoleic acid)c26.58 (4.21)27.04 (4.61)26.2 (3.81)<0.001
C18:3n-6 (gamma-linoleic acid)0.63 (0.17)0.21 (0.10)0.25 (0.12)<0.001
C20:2n-6 (eicosadienoic acid)0.18 (0.04)0.18 (0.04)0.17 (0.04)0.009
C20:3n-6 (dihomo-y-linoleic acid)1.04 (0.30)0.99 (0.30)1.08 (0.30)<0.001
C20:4n-6 (arachidonic acid)c3.86 (0.86)3.75 (0.83)3.95 (0.88)<0.001
n-3 PUFA6.52 (2.38)6.70 (2.49)6.37 (2.28)0.002
C18:3n-3 (linolenic acid)c0.63 (0.17)0.65 (0.19)0.61 (0.15)<0.001
C20:5n-3 (eicosapentaenoic acid)c2.05 (1.36)2.15 (1.45)1.97 (1.28)0.003
C22:5n-3 (docosapentaenoic acid)0.55 (0.15)0.56 (0.15)0.55 (0.14)0.087
C22:6n-3 (docosahexaenoic acid)3.29 (0.99)3.34 (1.01)3.24 (0.97)0.028
vlc n-3 PUFA5.89 (2.38)6.05 (2.49)5.76 (2.27)0.007
Plasma SCD indices (16:1n-7/16:0)0.09 (0.03)0.09 (0.02)0.10 (0.03)<0.001
Plasma SCD indices (18:1n-9/18:0)2.20 (0.35)2.20 (0.36)2.20 (0.35)0.67

We found significant correlations between dietary intake of PUFA (% of total energy) and plasma PUFA (% of total fatty acids). The strongest correlations were between intake of very long chain n-3 PUFA and plasma n-3 PUFA and 20:5n-3 (r = 0.40 and P < 0.001 for both), followed by the correlations between dietary and plasma n-6 PUFA (r = 0.38 and P < 0.001) and 18:2n-6 (r = 0.39 and P < 0.001).

Associations between plasma PUFA and plasma SCD indices

Linear regression analyses were used to investigate associations between plasma PUFA and plasma SCD indices (Table 2). We found strong inverse associations of total PUFA, total n-6 PUFA and total n-3 PUFA with both SCD indices after adjustment for sex and all other PUFA (Model 1). Among the individual PUFA, 18:2n-6, 20:4n-6, and 20:5n-3 were inversely associated with both SCD-16 and SCD-18 indices, whereas 18:3n-3 was inversely associated only with theSCD-16 index. The 18:2n-6 showed the strongest inverse association with both SCD-16 and SCD-18 indices followed by 20:5n-3.

Table 2. Associations between plasma concentrations of PUFA and plasma SCD indices (The Hordaland Health Study)
 Total population (n = 2,021)
 SCD-16SCD-18
 Partial rP valuePartial rP value
  1. Linear regression analysis using log-transformed SCD data. PUFA, polyunsaturated fatty acids; SCD, stearoyl-coenzyme A desaturase.

  2. a

    Adjusted for sex and all other PUFA.

  3. b

    Adjusted for sex, all other PUFA, triglycerides and cholesterol.

Model 1a
Total PUFA−0.65<0.001−0.63<0.001
n-3 PUFA−0.34<0.001−0.49<0.001
18:3n-3−0.15<0.0010.0040.85
20:5n-3−0.13<0.001−0.25<0.001
n-6 PUFA−0.65<0.001−0.61<0.001
18:2n-6−0.59<0.001−0.59<0.001
20:4n-6−0.14<0.001−0.060.010
Model 2b
Total PUFA−0.66<0.001−0.58<0.001
n-3 PUFA−0.39<0.001−0.46<0.001
18:3n-3−0.010.790.010.74
20:5n-3−0.20<0.001−0.24<0.001
n-6 PUFA−0.67<0.001−0.57<0.001
18:2n-6−0.57<0.001−0.53<0.001
20:4n-6−0.14<0.001−0.030.23

Controlling for sex, plasma concentration of TG and cholesterol (Table 2, Model 2), or sex, plasma concentration of TG and cholesterol, intake of alcohol, energy, and coffee, physical activity score, smoking, and time since last meal (Model 3, data not shown), only modestly changed the results. The strong inverse associations with plasma SCD-16 and SCD-18 indices remained for total PUFA, n-6 PUFA, n-3 PUFA, 18:2n-6, and 20:5n-3. The association of 20:4n-6 with SCD-16 index remained significant, whereas the relations between 18:3n-3 and plasma SCD-16 index and between 20:4n-6 and SCD-18 index were no longer significant. Additional adjustments for type and amount of dietary fat did not alter the associations.

To test whether the observed relations of plasma PUFA with SCD indices were driven by the relatively high plasma TG in our population [28], we examined the associations of PUFA with the indices according to quartiles of TG concentrations. The associations showed similar pattern in all TG quartiles (Supporting Information Table S2).

We also repeated the analysis stratified by fasting state to exclude the possibility that our results were confounded by prandial status. There results were similar in the nonfasting (1-6 h since last meal, n = 1952) and fasting (7-9 h since last meal, n = 50) groups. The associations of total PUFA, total n-6 PUFA, and n-3 PUFA with the plasma SCD-16 index were in fact stronger in the fasting group (Model 1: total PUFA and n-6 PUFA, partial r = −0.79; n-3 PUFA, partial r = −0.58, P < 0.001 for all) than in the nonfasting group (Model 1: total PUFA, partial r = −0.64; n-6 PUFA, partial r = −0.65; n-3 PUFA, partial r = −0.33, P < 0.001 for all). Similar results were also observed in Models 2 and 3 (data not shown)

Concentration-response curves were fitted to examine possible nonlinear relations between the plasma PUFA and SCD indices. The inverse associations of total PUFA, total n-6 PUFA, and n-3 PUFA with plasma SCD-16 and SCD-18 indices were strongly linear (Figure 1). The strongest linear dose-response relation of the individual PUFA with SCD-16 and SCD-18 indices, after adjustments for sex and all other PUFA, was for 18:2n-6, followed by 20:5n-3 (Figure 2), consistent with the linear regression results. The 18:3n-3 and 20:4n-6 were linearly but more weakly associated with plasma SCD-16 index, but not with SCD-18 index (Figure 2).

Figure 1.

Associations of total plasma concentration of PUFA, n-3 PUFA and n-6 PUFA (g/100g FAME) with plasma SCD indices. The curves were adjusted for sex. Additional adjustments were: C and D, n-6 plasma PUFA; E and F, n-3 PUFA. Solid lines represent the estimated dose response curves; shaded areas represent the 95% CI. The lowest and highest 2.5 percentiles of independent variables are not shown. Partial r and P values are from corresponding linear regression analyses.

Figure 2.

Associations of plasma concentrations of 18:2n-6, 20:4n-6, 18:3n-3, and 20:5n-3 (g/100g FAME) with plasma SCD indices. The curves were adjusted for sex and all other PUFA. Solid lines represent the estimated dose response curves; shaded areas represent the 95% CI. The lowest and highest 2.5 percentiles of independent variables are not shown. Partial r and P values are from corresponding linear regression analyses.

Associations between plasma PUFA and fat mass

Linear regression and generalized additive models were used to investigate associations between plasma PUFA concentrations and fat mass (Figure 3). We found significant inverse associations of total plasma PUFA, n-6 PUFA and n-3 PUFA with fat mass. Adjusting for SCD-16 markedly reduced the strength of the associations. From the 2.5th to 97.5th percentile of the plasma total n-6 PUFA concentration, fat mass differed by about 7 kg (adjusted for sex, lean mass and n-3 PUFA). After including SCD-16 into model, the difference was 2.5 kg.

Figure 3.

Associations of total plasma concentration of PUFA, n-3 PUFA and n-6 PUFA (g/100g FAME) with fat mass (kg). The curves were adjusted for sex and lean mass (A-C) and sex, lean mass, and SCD-16 index (D-F). Additional adjustments were: B and E, n-6 PUFA; C and F, n-3 PUFA. Solid lines represent the estimated dose response curves; shaded areas represent the 95% CI. The lowest and highest 2.5 percentiles of independent variables are not shown. Partial r and P values are from corresponding linear regression analyses.

Discussion

In this large-population based study, we found that plasma PUFA, especially n-6 PUFA, were inversely associated with body fat mass measured by DXA, and with plasma SCD indices. The relationship of PUFA with fat mass appeared to be partly mediated via SCD, since the association weakened after adjusting for SCD indices. Among the individual PUFA, 18:2n-6 showed the strongest associations with SCD indices and fat mass, indicating that n-6 PUFA could be an important factor in SCD regulation.

We have previously reported strong inverse associations between dietary intake of PUFA (total PUFA, n-6 and n-3 PUFA) and plasma SCD indices in this study population [8]. Some plasma fatty acids can be used as biomarkers of dietary fat [29, 30]. Consistent with earlier reports [30], we found positive correlations between dietary PUFA and plasma PUFA, especially between dietary n-3 PUFA and plasma 20:5n-3, and dietary n-6 PUFA and plasma 18:2n-6. Because there are limitations related to the collection of dietary data, the use of fatty acid composition in plasma provides an objective biomarker of the quality of dietary fat intake [29].

The main finding in this study is that high levels of plasma PUFA are linearly and strongly associated with lower SCD indices. To our knowledge, this is the largest study that has examined associations between individual plasma PUFA and SCD indices. Small clinical crossover studies have investigated the effects of different PUFA diets on SCD indices [31-33]. Velliquette et al [31] reported a negative association of serum 18:2n-6, but a positive association of serum 18:3n-3 with serum SCD indices in their cohort (n = 23). Another controlled crossover study of men and women (n = 20) found a significantly higher SCD index (16:1n-7/16:0) in serum cholesteryl esters and phospholipids in subjects consuming a SFA-rich diet compared to the rapeseed oil diet (rich in 18:1n-9, 18:2n-6 and 18:3n-3) [33]. Furthermore, a strong negative correlation (r = −0.7, P < 0.001) between marine n-3 fatty acids (20:5n-3 and 22:6n-3) and SCD-18 index in plasma phospholipids was demonstrated in men and women (n = 30) consuming either 150 g of cod, salmon, or potato (control) daily for 15 days [32]. In contrast to the sparse data in humans, it is well known that PUFA-rich diets in rodents repress several lipogenic genes, including SCD1, resulting in reduced levels of MUFA, TG, and cholesteryl esters in plasma and liver [10, 21].

We found that both n-6 PUFA, in particular 18:2n-6 and the n-3 PUFA, mainly 20:5n-3 were strongly inversely associated with plasma SCD indices. In line with these findings, a recent randomized control trial reports that n-6 PUFA feeding reduced SCD-16 index compared with a SFA diet [34]. In Hep G2 cells, 18:2n-6 and 20:5n-3, but not 18:3n-3, decreased SCD mRNA, via SREBP-1c [31]. The authors suggested that 18:3n-3 may not regulate SCD directly, but via its bioconversion to 20:5n-3 [31]. Dietary 18:2n-6 and 18:3n-3 may be metabolized to the end products of PUFA synthesis, that is, 20:4n-6 and 22:6n-3, respectively. However, the conversion of dietary 18:3n-3 into 20:5n-3 and further to 22:6n-3 is limited [35].

Interestingly, n-6 PUFA appeared more strongly associated with SCD indices than n-3 PUFA. A potential explanation is that 18:2n-6 may reduce synthesis of 20:5n-3 due to competition for common enzymes responsible for desaturation and elongation of PUFA [12]. Thus, a high intake of 18:2n-6 may disturb the metabolism and distribution of n-3 PUFA. Notably, 18:2n-6 is the predominant PUFA in the Western diet [12]. The 18:2n-6 is abundant in vegetable oils, seeds, and nuts, whereas n-3 PUFA is mainly obtained from fatty fish, cod liver oil and fish oils [12]. Our study participants had a relatively high intake of n-3 PUFA from fish and fish oils [26]. Despite such high intake, we observed no interaction between n-3 or n-6 PUFA, and the inverse dose-response relation with SCD indices was linear from low to high concentrations of total PUFA.

We observed that within the normal range of PUFA concentrations, which are at least partly determined by PUFA intake, SCD indices are markedly reduced as PUFA concentrations increase. From low to high plasma PUFA concentration, SCD-16 index decreased by about 0.07. Interestingly, this change in plasma SCD-16 index is associated with a 6-8 kg difference in fat mass in the same population [8]. Is it possible that PUFA might have such an effect on body composition? In the present study, plasma total PUFA, n-6 and n-3 PUFA were inversely associated with fat mass; with a fat mass difference of about 7 kg from low to high PUFA levels. This association was attenuated after adjusting for SCD-16 index, suggesting that increasing plasma PUFA may decrease body fat partly via their effect on SCD activity. There are some small randomized control trials suggesting that in obese subjects n-3 [36] and n-6 [34] PUFA reduce adipose tissue inflammation and liver fat, respectively.

Beyond the mechanisms already described, metabolic studies in humans and rodents have shown that high PUFA levels may favor fat oxidation in several tissues including adipose, liver, cardiac, intestinal and skeletal muscle tissue and reduced fat deposition in some adipose tissues [15]. Experimental studies also show that PUFA suppression of SCD1 contributes to the inhibition of de novo lipogenesis [9, 11]. Thus, the potential benefit of PUFA in obesity is the change in metabolism toward enhanced fat oxidation and reduced fatty acid synthesis and storage [9]. Although both n-6 and n-3 PUFA were inversely associated with plasma SCD indices and fat mass, our findings points to a potentially greater importance of n-6 PUFA in lipid metabolism and possibly body fat regulation. Although some epidemiological studies and animal models may suggest that n-6 PUFA promote adiposity [18, 19], controlled studies show that diets rich in n-6 PUFA reduced subcutaneous adipose tissue [37] and trunk adipose tissue [38]. Thus, more studies carefully designed are required to examine the role of different dietary PUFA in human energy balance.

The strengths of the present study include the large sample size recruited among community-dwelling older adults. The narrow age range of the study subjects (71-74 years) avoids confounding due to age, although our findings cannot be generalized to other age groups. We used plasma SCD indices calculated from fatty acid profiles of total plasma lipids as indirect markers of SCD1 enzyme activity. Using plasma total lipids rather than defined lipid fractions may be a weakness of our study [4, 7]. However, in a dietary animal model, SCD-16 index in total plasma lipids tracked changes in SCD1 expression in liver, but not adipose tissue [39]. Furthermore, consistent with our findings, Crowe et al [40] reported an inverse association of PUFA intake with plasma 16:1n-7 measured in the separate lipid fractions.

Another limitation is the non-fasting measurements [30], although adjusting the analyses for time since last meal, plasma concentration of TG and intakes of energy, and fat did not alter the results. In fact, the nonfasting measurements may have diluted the association between plasma PUFA and SCD indices, since we observed stronger, significant associations in the small fasting subgroup. We have interpreted our data to indicate that high PUFA may lower fat mass via an influence on SCD activity, since animal experiments and human trials show SCD suppression in response to PUFA-rich diets [13, 21, 31-33]. Finally, the cross-sectional design of the study precludes conclusive inferences on causality and cannot exclude confounding or reverse causality.

In conclusion, we observed that plasma PUFA, in particular n-6 PUFA, show strong inverse associations with SCD indices and fat mass. These findings are in accordance with animal studies and human trials showing suppression of SCD1 expression or activity, and improved body composition in response to PUFA-rich diets. Our data highlights the importance of n-6 PUFA, which so far have received less attention than n-3 PUFA in relation to body fat regulation. Given that SCD activity is strongly associated with adiposity, further studies should be carried out to investigate whether dietary PUFA may influence body weight.

Ancillary