A, Correlations between proportions of fatty acids in adipose tissue (weight percentage [wt%]) and dietary records (percentage of fat) in all men and in men not underreporting dietary intake according to the Goldberg cutoff as modified by Black15 (approximately 53%). B, Correlations between proportions of fatty acids in serum cholesterol esters (wt%) and dietary records (percentage of fat) in all men and in adequate reporters (approximately 53%). C, Correlations between proportions of fatty acids in adipose tissue (wt%) and cholesterol esters (wt%). D, Correlations between adipose tissue palmitic (16:0) and palmitoleic (16:1n-7) acids (wt%) and reported dietary intakes (% of energy) in all men. SFA, saturated fatty acids. CHO, carbohydrate.
aP ≤ .001.
bP < .01.
cP < .05.
Unadjusted data (n = 853). The solid blue line denotes relative hazard ratio (HR), the dashed blue lines represent upper and lower limits for 95% CIs; the dotted light blue line represents reference value. CHO indicates carbohydrate; SFA, saturated fatty acids.
eFigure 1. Study population
eTable 1. Adipose tissue arachidonic acid and all-cause mortality, adjusted for different covariates
eTable 2. Number (and %) of deaths by levels of linoleic acid (%)
eFigure 2. Restricted cubic spline for the association between adipose tissue arachidonic acid and all-cause mortality
Customize your JAMA Network experience by selecting one or more topics from the list below.
Iggman D, Ärnlöv J, Cederholm T, Risérus U. Association of Adipose Tissue Fatty Acids With Cardiovascular and All-Cause Mortality in Elderly Men. JAMA Cardiol. 2016;1(7):745–753. doi:10.1001/jamacardio.2016.2259
Are intakes of polyunsaturated fatty acids (as reflected in adipose tissue) associated with cardiovascular and all-cause mortality?
In a Swedish community-based cohort study of 853 men, 71 years old, adipose tissue linoleic acid was associated with lower all-cause mortality but not cardiovascular mortality over a 15-year follow-up.
Omega-6 intake was associated with lower risk of death in elderly men.
The major polyunsaturated fatty acids in adipose tissue objectively reflect long-term dietary intake, and may provide more reliable information than would self-reported intake. Whether adipose tissue fatty acids predict cardiovascular and all-cause mortality needs investigation.
To investigate associations between adipose tissue fatty acids and cardiovascular and overall mortality in a cohort of elderly men.
Design, Setting, and Participants
We hypothesized that polyunsaturated fatty acids reflecting dietary intake, are inversely associated with cardiovascular and all-cause mortality. In the Swedish cohort study Uppsala Longitudinal Cohort of Adult Men, buttock fatty acid composition was analyzed by gas-liquid chromatography in 1992 to 1993 and 2008. The study participants were followed during 11 311 person-years, between 1991 and 2011 (median follow-up, 14.8 years). In this community-based study that took place from 1970 to 1973, all men born in 1920 to 1924 in Uppsala, Sweden, were invited and 2322 (82%) were included (at age 50 years). At the reinvestigation at age 71 years, 1221 (73%) of the 1681 invited men participated. Adipose tissue biopsy specimens were taken in a subsample of 853 men. There was no loss to follow-up.
Adipose tissue proportions of 4 polyunsaturated fatty acids that were considered to mainly reflect dietary intake (linoleic acid, 18:2n-6; α-linolenic acid, 18:3n-3; eicosapentaenoic acid, 20:5n-3; and docosahexaenoic acid, 22:6n-3) comprised primary analyses, and all other available fatty acids were secondary analyses.
Main Outcomes and Measures
Hazard ratios (HRs) for cardiovascular and all-cause mortality using Cox proportional hazards regression analyses, performed in 2015.
Among the 853 Swedish men, there were 605 deaths, of which 251 were cardiovascular deaths. After adjusting for risk factors, none of the 4 primary fatty acids were associated with cardiovascular mortality (HR, 0.92-1.05 for each standard deviation increase; P ≥ .27). Linoleic acid was inversely associated with all-cause mortality (HR, 0.90; 95% CI, 0.82-0.98; P = .02) and directly associated with intake (P < .001). In secondary analyses, palmitoleic acid, 16:1n-7 (HR, 1.11; 95% CI, 1.02-1.21; P = .02) was associated with higher all-cause mortality, whereas heptadecanoic acid, 17:0, tended to be associated with lower all-cause mortality (HR, 0.89; 95% CI, 0.79-1.00; P = .05). Arachidonic:linoleic acid ratio was associated with both cardiovascular (HR, 1.15; 95% CI, 1.05-1.31; P = .04) and all-cause (HR, 1.13; 95% CI, 1.04-1.23; P = .005) mortality.
Conclusions and Relevance
Adipose tissue linoleic acid was inversely associated with all-cause mortality in elderly men, although not significantly with cardiovascular mortality.
Dietary recommendations emphasize high intakes of polyunsaturated fatty acids (PUFAs) over saturated fatty acids (SFAs), that is, less fat from animal sources and more from nontropical vegetable oils, nuts, and fish.1 The scientific evidence for these recommendations is based on both prospective observational studies2 and a limited number of interventions with cardiovascular disease (CVD) end points.3 Also, a large body of evidence exists regarding the short-term effects of different dietary fatty acids on blood lipoproteins,4 demonstrating cholesterol-lowering effects of PUFAs compared with SFAs, although the mechanisms are not entirely clear.5 In all, institutions such as the Food and Agriculture Organization of the United Nations and the World Health Organization have until recently regarded the evidence for the benefits of replacing SFAs with PUFAs to fight CVD as convincing.6 However, some controversy remains regarding the role of the predominant dietary PUFAs, linoleic acid (LA) (18:2n-6), which has been suspected to increase inflammation through endogenous conversion to arachidonic acid (20:4n-6), a precursor of both pro-inflammatory and anti-inflammatory eicosanoids. Also, LA increased cardiovascular morbidity and all-cause mortality at high intakes of a study margarine in reanalyses of the Sydney Diet Heart Study.7 To date, no human trials, however, have demonstrated that moderate amounts of dietary LA increase markers of inflammation.8-10 In contrast, a recent systematic review11 of prospective observational studies concluded that dietary LA was inversely associated with coronary artery disease risk and mortality in a dose-response manner. However, another meta-analysis12 reported no associations between dietary, circulating, or tissue n-6 PUFA, and coronary events, although mortality was not reported.
Because there are only few randomized clinical trials on dietary PUFAs and mortality, prospective observational studies with long-term follow-up are of high importance. However, measurement errors and underreporting of dietary intake of fats lower the reliability of these studies. An alternative approach is using objective markers of dietary intake instead, that is, adipose tissue composition, which for certain fatty acids reflects long-term dietary intake reasonably well.13 Because very few prospective studies have examined the association between adipose tissue fatty acids and mortality outcomes, such studies are highly warranted. Also, to our knowledge, no such studies in elderly populations, in whom mortality is high, have been published. Previous studies in the present cohort have used serum cholesterol esters as fatty acid biomarkers and have found inverse associations between LA and cardiovascular and total mortality in 50-year-old men,14 but no such data exist on adipose tissue PUFA or in elderly Swedish men. We hypothesized that adipose tissue PUFA (LA, α-linolenic [ALA]; eicosapentaenoic [EPA]; and docosahexaenoid [DHA] acids) would reflect dietary intakes, and would be associated with lower cardiovascular and all-cause mortality in a population-based cohort of elderly Swedish men.
The Uppsala Longitudinal Study of Adult Men (ULSAM) is a cohort study conducted in Uppsala, Sweden. Only men were included because the study’s main purpose was to investigate causes of CVD in middle-aged men, in whom cardiovascular mortality was high in the 1970s. The present study originated from ULSAM’s third examination in 1991 to 1995, when participants were approximately 71 years old. The study was approved by the Interdisciplinary Collaboration Team on Uppsala Longitudinal Studies at the Section of Geriatrics at Uppsala University and by the Ethics Committee of Uppsala University. Participants provided written informed consent; they were not compensated for taking part in the study. Up to this point, 422 participants had died and 219 moved out of the region. Of the 1681 invited men, 1221 chose to participate at this point. At the time of the analysis in 2015, complete data from death registries through December 31, 2011, were available. See eFigures 1 and 2, and eTables 1 and 2 in the online Supplement for additional information.
Participants’ medical history, weight, height, body mass index (BMI), and waist circumference were obtained at the baseline visit at age 71 years, in 1991 to 1995. Blood pressure was measured in the supine position. Questions about current smoking status (smoker vs nonsmoker) and education level (elementary school, secondary school, or university studies) were asked by a study nurse. Biochemical analyses were performed using routine laboratory analyses, and low-density lipoprotein cholesterol (LDL-C) level was calculated by the Friedewald formula. Dyslipidemia was defined as taking lipid-lowering medication or having an LDL-C level greater than 5 mg/dL or high-density cholesterol (HDL-C) level less than 38.6 mg/dL (to convert LDL-C and HDL-C to millimoles per liter, multiply by 0.0259). Physical activity was assessed by an optically readable questionnaire, graded 1 (sedentary) to 4 (hard physical training). Dietary fat and alcohol intakes were determined from a 7-day optically readable food diary, and fatty acids were given as weight percentage (wt%) of total fat intake. Adequate reporters were determined by the Goldberg cutoff, as modified by Black15 for acceptable reported energy intake:basal metabolic rate (BMR) at the individual level, that is, taking into account the number of days reported (7) and within-subject variations in energy intake (23%), BMR (8%), and variation in physical activity (15%). Physical activity levels were estimated as 1.4, 1.5, 1.6, and 1.7 times BMR, for each reported activity grade. The BMR was determined by the Harris-Benedict equation.16
Information on dates and causes of death were collected from the Swedish Cause of Death registry for all participants every calendar year. International Classification of Diseases, Ninth Revision (ICD-9), and International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10), codes 390 to 459 and I00 to 99 were used, respectively, for classification of deaths as cardiovascular mortality.
In the first 853 men examined (1991-1995), adipose tissue biopsy specimens were taken from the outer quadrant of the buttocks with a needle coupled to a vacuum tube. The sample was collected in the connector between the needle and the tube, and stored at −70°C until analysis. Prior to the analysis, the sample was weighed and homogenized. Fatty acid composition was analyzed by gas-liquid chromatography according to Boberg et al,17 using a Hewlett Packard gas-liquid chromatograph system, 7671A autosampler, 3392A integrator, and 25-m Quadrex Fused Silica capillary column OV-351, with helium as the carrier gas. The amounts of fatty acids are given as relative percentages of the sum of all fatty acids (as wt%).18 The fatty acids were analyzed in 2 batches; on the first occasion (approximately 1992-1993; n = 318), the fatty acids 16:0, 16:1n-7, 18:0, 18:1n-9, 18:2n-6, 18:3n-3, 18:3n-6, 20:3n-6, 20:4n-6, 20:5n-3, 22:4n-6, 22:5n-3, and 22:6n-3 were included. Fatty acid composition of serum cholesterol esters was analyzed using similar methods in a subsample of 611 men. In 2008, that is, 15 to 16 years later, the remaining 535 adipose tissue samples were analyzed in the same laboratory using an Agilent Technologies system consisting of a model 6890N gas-liquid chromatograph, 7683 autosampler, and an Agilent ChemStation; this time, the minor fatty acids 12:0, 14:0, 15:0, and 17:0 were also included, and thus slightly different methods were used, that is, higher temperature and a 30-m glass capillary column coated with Thermo TR-FAME (Thermo Electron Corp).18 Steaoryl-CoA desaturase (SCD) activity in adipose tissue was estimated as product-to-precursor ratios of individual fatty acids (ie, 16:1n-7/16:0 and 18:1n-9/18:0). Conversion from 18:2n-6 to 20:4n-6 was estimated as the ratio 20:4n-6/18:2n-6.
Cox proportional hazards models were used to estimate HR for each fatty acid, standardized at a 1 SD increase. Results are presented for 2 separate models, 1 crude model adjusting for age and analysis occasion (0 of 1), and a second model adjusting also for CVD risk factors, including smoking (data were available for 844 men), BMI (848 men), alcohol intake (789 men), physical activity (791 men), diabetes prevalence (852 men), systolic blood pressure (851 men), dyslipidemia (853 men), and hypertension treatment (849 men). These 2 models were selected after drawing causal diagrams for the most important associations and investigator discussions. Data are given as the mean (SD) for all variables. Normality of variables was determined by the Shapiro-Wilk test and visual examinations of histograms. Adipose tissue LA, ALA, EPA, DHA, 18:3n-6, 20:3n-6, 22:4n-6, 12:0, and alcohol intake were considered nonnormally distributed and thus natural log-transformed before analyses. Ten iterations of multiple imputation of missing data on covariates were performed (STATA command ice), without any imputation of data on exposure variables. Pairwise correlations were analyzed as Pearson r, granted at least 1 variable was approximately normally distributed. Significant associations in both models among primary analyses were further investigated using restricted cubic splines (without a set reference point; STATA command uvrs). P < .05 was considered statistically significant. All analyses were performed with STATA statistical software (versions 11.0 and 13.0; StataCorp).
Baseline characteristics and adipose tissue fatty acid composition at age 71 years are presented in Table 1. Most adipose tissue fatty acids were correlated with dietary record data (as percentage of fat intake) (Figure 1A); r = 0.37 for LA, r = 0.16 for ALA, r = 0.25 for EPA, and r = 0.32 for DHA; r = 0.20 for 16:0, r = 0.38 for 14:0, and r = 0.15 for 12:0; P < .001 for all comparisons. Neither 18:0 (r = 0.03; P = .37) nor 18:1n-9 (r = 0.01; P = .89) were, however, correlated with dietary records, and 20:4n-6 was only weakly correlated with dietary records (r = 0.08; P = .02). In adequate reporters (n = 419 for LA, that is, 53% of the 787 individuals with available data from both adipose tissue and dietary records), associations were even stronger for most PUFAs and 16:0, where r = 0.47 for LA, r = 0.27 for ALA, r = 0.29 for EPA, r = 0.36 for DHA, and r = 0.23 for 16:0 (P < .001 for all comparisons). Other fatty acids were less different in adequate reporters; r = 0.13 (P = .03) for 12:0, r = 0.38 (P < .001) for 14:0, r = 0.16 (P = .001) for 16:1n-7, r = 0.03 (P = .51) for 18:0, r = −0.00 (P = .98) for 18:1n-9, and r = 0.10 (P = .05) for 20:4n-6 (Figure 1B). All available fatty acids in adipose tissue and serum cholesterol esters were also strongly (P < .001) correlated in these 2 compartments, particularly LA (r = 0.61) (Figure 1C). Both 16:0 and 16:1n-7 were associated with reported SFA intakes and negatively associated with carbohydrate intakes (Figure 1D). No association was found between self-reported dietary LA intake and the proportions of arachidonic acid (AA) in adipose tissue (data not shown).
The 853 participants were followed for a median of 14.8 years (range, 0.1-20.2 years; 11 311 total person-years). There were 605 deaths, and 251 of these were cardiovascular deaths. After adjustment for CVD risk factors, there were no significant associations with cardiovascular mortality (Table 2). In the minimally adjusted model, stearic acid (18:0) was associated with decreased cardiovascular mortality, whereas the very long chain n-6 PUFA arachidonic acid (20:4n-6) and 22:4n-6 were associated with increased cardiovascular mortality.
For all-cause mortality, LA was associated with decreased mortality in both models (Table 3). In the risk factor–adjusted model, palmitoleic acid (16:1n-7) was associated with increased mortality, whereas heptadecanoic acid (17:0) was inversely associated with mortality (secondary analyses). Stearic acid (18:0), arachidonic acid (20:4n-6), and 22:4n-6 were associated with decreased mortality in the age-adjusted model, but these associations disappeared after adjusting for CVD risk factors (Table 3). The association with mortality for AA was not explained by any single covariate (eTable 1 in the Supplement). The number of deaths at different levels of LA is given in eTable 2 in the Supplement.
To test that the association with LA and mortality was independent of socioeconomic status, length of education was also added as a covariate to the CVD risk factor model, and new multiple imputations were performed, including education. However, the results were not affected; for CVD mortality, hazard ratios (HRs) and 95% CIs were 0.92 (95% CI, 0.80-1.06; P = .27) both including and excluding education. For all-cause mortality, HRs were 0.90 (95% CI, 0.82-0.99) in both models (P = .02 excluding education, and P = .03 including education).
In a restricted cubic spline analysis, the inverse association between LA and mortality demonstrated a clear association for higher mortality at the lowest LA levels, with a tendency toward still lower HR also at the highest levels (Figure 2).
In this prospective cohort study of elderly men with 15 years of follow-up and 605 deaths, adipose tissue LA, the major dietary n-6 PUFA, was associated with lower all-cause mortality, with a tendency toward lower CVD mortality. Considering that LA is an essential fatty acid and one of the best biomarkers for n-6 PUFA intake,13 the results may suggest a beneficial role of PUFA-rich vegetable oils. Such a possibility is further supported by a current strong association between LA intake and adipose tissue LA percentage. To our knowledge, this is the largest study presenting such data. In secondary analyses, the monounsaturated fatty acid (MUFA) palmitoleic (16:1n-7) was associated with increased mortality, whereas heptadecanoic acid (17:0) was associated with decreased mortality.
Few similar prospective studies have investigated associations between adipose tissue and CVD, and we are not aware of any that have investigated cardiovascular or all-cause mortality. In the Scottish Heart Health Extended Cohort Study, upper arm adipose tissue fatty acids were sampled in 3944 middle-aged men and women who were followed for a median of 19.5 years.19 The n-3 PUFA 20:5n-3 and 22:6n-3 were associated with lower risk of myocardial infarction (MI), a finding that we could not observe with regard to cardiovascular mortality. The effects of dietary n-3 PUFA may differ in middle-aged vs elderly populations. For LA, there was a trend for decreased risk of MI, but this was not significant after adjusting for various lifestyle variables; all results were also adjusted for total and HDL-C levels in the Scottish study. We instead chose to adjust only for the categorical factor dyslipidemia, to capture potential confounding by medical treatment, since protective effects of LA and ALA may be mediated by LDL-C level lowering. Although LA was associated with decreased all-cause mortality, we could not, however, clearly confirm any potentially protective effects against CVD risk for any PUFA. This may have been due to the relatively few CVD events, as there was a tendency for lower CVD mortality for LA. In the Scottish study, but not ours, both palmitoleic acid (16:1n-7) and oleic acid (18:1n-9) (ie, MUFA) were associated with increased risk of MI. Palmitoleic acid was, however, congruently associated with higher all-cause mortality in the present study. A higher mortality risk at higher adipose tissue proportions of MUFA might reflect increased SCD-1 activity, since this enzyme produces MUFA from SFAs, and palmitoleic acid is uncommon in the diet. Such reasoning is supported by the current significant association between increased adipose tissue SCD-1 activity and mortality, in line with previous findings using hepatic SCD-1 activity indices in this population.14 Stearic acid (18:0) tended to be associated with decreased CVD risk, in line with our study. This, however, probably reflects factors other than diet (eg, elongation from palmitic acid), since we found no correlation between adipose tissue 18:0 and dietary records. Apart from other SFAs, stearic acid does not increase LDL-C level.20 In secondary analyses we also observed an inverse association between the milk fat biomarker, 17:0, and all-cause mortality. This may suggest a beneficial role of milk fat intake or of dairy components (eg, calcium, magnesium, vitamin D). An inverse link between dairy intake and all-cause mortality has, however, not been shown in previous prospective cohort studies,21 and supportive evidence from randomized clinical trials is lacking. Until such evidence or mechanisms are available, residual confounding must be considered. Because adipose tissue palmitic acid is the main product of de novo lipogenesis in positive energy balance, we investigated whether palmitic acid was associated with reported dietary intakes of SFAs and carbohydrate. There were positive correlations with SFAs and 16:0 intakes and an inverse correlation with carbohydrate intake, in line with the notion that in populations with a relatively high fat intake, most adipose tissue palmitic acid is probably derived from SFA intake rather than from carbohydrates.22 The association between 16:1n-7 and SFA intake accord with controlled interventional data showing increased 16:1n-7 in serum lipids in response to increased SFA intake and decreasing MUFA and PUFA intake.23
The association between adipose tissue AA and increased risk of CVD mortality, in the age-adjusted models, is not consistent with serum AA of this cohort but may reflect altered desaturation and elongation of LA,14 because genetic metabolic and factors may mainly determine AA tissue levels rather than its dietary intake.24,25 Interestingly, a higher AA:LA ratio, potentially reflecting higher conversion rate from LA to AA, was associated with higher risk of both CVD and all-cause mortality. Whether this finding mainly reflects a genetic influence on this ratio remains to be investigated, but recently a variant in the FADS1 gene was suggested to alter the LA to AA conversion rate in erythrocyte membrane and plasma phospholipids in elderly Japanese patients.26
One case-cohort study of middle-aged Danish men reported results on acute coronary syndrome (ACS)27 and MI28 for gluteal adipose tissue fatty acids. During a median follow-up of 7.6 years, there were 1012 cases of ACS. Both individual very long chain n-3 (22:6n-3 and docosapentaenoic acid, 22:5n-3 but not 20:5n-3) and overall n-3 PUFA were inversely associated with ACS risk compared with 1630 healthy controls from the cohort. This was observed in men but not women, and most cases (86%) were nonfatal, which may explain the inconsistency with the present study, in which we found no associations between n-3 PUFA and cardiovascular mortality. A study of 2134 cases with MI reported an increased risk associated with adipose tissue arachidonic acid (associations with other fatty acids were not reported), which was also not believed to be reflective of dietary intake, because adipose tissue arachidonic acid was not associated with dietary arachidonic acid or LA.28 A similar tendency for increased CVD mortality was observed in the present study for minor very long chain n-6 fatty acids, and the absence of a clear association might have been due to insufficient statistical power.
There are also several retrospective studies on adipose tissue fatty acids and cardiovascular outcomes29-48 or metabolic syndrome,49 with divergent results. Such biomarker studies do not rely on a patient’s memory, but incident CVD may result in an individual’s changed dietary habits, which may introduce bias through reverse causation. We will therefore avoid discussing retrospective studies in further detail.
The remaining causes of deaths were unknown, but it is possible that dietary LA may be inversely linked to cancer mortality. Mechanisms for such associations are, however, unknown, and more studies are required to confirm the potential of LA in promoting longevity in elderly populations. Of note is that during the early 1990s, common margarines contained both LA and trans fatty acids, thus possibly a negative confounding factor in these associations. Indeed, unpublished observations of a subsample of 102 men of the present cohort demonstrated direct associations between adipose tissue LA and trans fatty acids derived from partially hydrogenated vegetable oils (oral communication; Bengt Vessby, MD, PhD; December 2014). Considering that we show novel data that the relative levels of LA in adipose tissue is associated with that of serum LA, it is noteworthy that our results accord with those of a recent cohort study50 of 60-year-olds, showing inverse relations between serum cholesterol esters and all-cause mortality. In that study also, LA showed inverse but nonsignificant associations with CVD events in adjusted models.
Of note, the associations between adipose tissue and reported dietary intake were highly correlated for certain fatty acids (eg, LA). The association was stronger in men who did not underreport dietary intakes. These findings indicate that adipose tissue seems to be a valid objective marker of PUFA intake, especially LA that is present in higher proportions in adipose tissue and has high measurement accuracy (coefficient of variation, 1%-2%) compared with other PUFA. Serum cholesterol esters fatty acids were also correlated with intake (except for 18:3n-3), but less strongly. This indicates that, when possible to ascertain in cohort studies, adipose tissue may be preferable to blood samples as biomarker for long-term dietary intake.
The strengths of the study were its population-based design with a high rate of participation and virtually no loss to follow-up, owing to reliable registry data. The follow-up period of 15 years was probably sufficient to determine the effects of baseline dietary factors. The study was unique in using adipose tissue as an objective marker of fatty acid intake for mortality outcomes and adds to the knowledge from previous studies in younger populations. Limitations include the observational study design with potential residual confounding, and it should therefore be noted that no causality can be implied by our study findings. Furthermore, there were relatively few cases with CVD deaths and thus limited power to detect strong associations with fatty acids. Similarly, for some minor SFAs (eg, 12:0 and 14:0) power was even lower owing to detection problems of these SFAs in adipose tissue. For fatty acids with low concentrations (eg, EPA), the measurement error was larger (CV was approximately 5%), which may have biased those results toward the null. Also, there were multiple primary analyses performed, and we cannot completely exclude the possibility that some associations are by chance. The risk of CVD-related death may also rely more on dietary intakes earlier in life, which adipose tissue at age 71 years may not adequately reflect. Also, the adipose tissue analyses were performed at 2 occasions with the tissue in a freezer for a long period of time in between and using slightly different methods. Including the analysis occasion as a categorical variable in the regression analyses, however, changed the results only slightly (toward stronger results). Furthermore, an interaction analysis in a previous study18 of the samples did not find any evidence for a freezer effect. The cohort consisted of only men, and results may not be generalizable to women or to populations with n-6 PUFA intakes higher than approximately 10% of energy as roughly estimated from food diaries.51
Using this rough estimation further, a LA intake of less than approximately 4% of energy in this population seemed to have increased mortality. Specific dietary recommendations cannot be given, because no randomized trials have linked dietary modification of adipose tissue fatty acid composition with outcomes.
We found no associations between fatty acid composition and cardiovascular mortality for adipose tissue PUFAs in 71-year-old men during 15 years of follow-up. However, adipose tissue LA, correlated with dietary LA intake, was independently associated with decreased all-cause mortality. Further studies should investigate the role of dietary PUFAs as well as adipose tissue desaturation in longevity.
Corresponding Author: David Iggman, MD, PhD, Unit for Clinical Nutrition and Metabolism, Uppsala Science Park, 75185 Uppsala, Sweden (email@example.com).
Accepted for Publication: May 31, 2016.
Published Online: August 17, 2016. doi:10.1001/jamacardio.2016.2259
Author Contributions: Dr Iggman had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: All authors.
Acquisition, analysis, or interpretation of data: Iggman, Ärnlöv, Risérus.
Drafting of the manuscript: Iggman.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Iggman.
Obtained funding: Cederholm.
Administrative, technical, or material support: Cederholm.
Study supervision: Ärnlöv, Risérus.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Cederholm has received grants from Nutricia, Nestlé, and Fresenius-Kabi. No other disclosures are reported.
Additional Contributions: We thank Siv Tengblad, BSc, for excellent laboratory work and Matti Marklund, PhD, for assistance in data editing. Both are from the Department of Public Health and Caring Sciences, Uppsala University. They were not compensated for their contributions.