Escolar Documentos
Profissional Documentos
Cultura Documentos
IMPORTANCE The major polyunsaturated fatty acids in adipose tissue objectively reflect
Supplemental content at
jamacardiology.com
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.
OBJECTIVE 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.
EXPOSURES Adipose tissue proportions of 4 polyunsaturated fatty acids that were
considered to mainly reflect dietary intake (linoleic acid, 18:2n-6; -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.
RESULTS 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.
(Reprinted) E1
Methods
Study Population
The Uppsala Longitudinal Study of Adult Men (ULSAM) is a cohort study conducted in Uppsala, Sweden. Only men were inE2
Key Points
Question Are intakes of polyunsaturated fatty acids (as reflected
in adipose tissue) associated with cardiovascular and all-cause
mortality?
Findings 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.
Meaning Omega-6 intake was associated with lower risk of death
in elderly men.
Baseline Measurements
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 (LDLC) 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 withinsubject 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
jamacardiology.com
Mean (SD)
71.0 (0.6)
Weight, kg
81.0 (11.7)
26.5 (3.5)
Waist circumference, cm
Systolic blood pressure, mm Hg
Diastolic blood pressure, mm Hg
95.1 (9.7)
147.2 (18.6)
84.3 (9.2)
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 70C 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.
151.0 (35.0)
168 (19.7)
Statistical Analysis
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 be-
226.0 (39.0)
249 (29.2)
128 (15.0)
12.7 (2.8)
1.0 (0.2)
0.1 (0.1)
0.3 (0.1)
0.5 (0.2)
3.4 (0.6)
0.3 (0.1)
21.9 (2.0)
6.8 (1.8)
0.2 (0.04)
4.0 (1.0)
49.3 (2.3)
0.1 (0.04)
0.2 (0.06)
0.4 (0.1)
0.1 (0.04)
0.3 (0.1)
fore 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).
Results
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,
jamacardiology.com
E3
Figure 1. Correlations Between Fatty Acid Composition in Adipose Tissue in 853 Samples and Serum
in 433 Samples, and Self-Reported Dietary Intake in 789 Samples
A Adipose tissue and self-reported intake
0.5
a
All men
0.4
Adequate reporters
Pearson r
a
0.3
a
a
a
a
0.2
b
c
0.1
0
18:2n-6
18:3n-3
20:5n-3
22:6n-3
12:0
14:0
16:0
16:1n-7
18:0
18:1n-9
20:4n-6
Fatty Acid
B
0.5
Adequate reporters
0.4
Pearson r
0.3
a
a
0.2
a
c
0.1
0
0.1
b
0.2
18:2n-6
18:3n-3
20:5n-3
22:6n-3
16:0
16:1n-7
18:0
18:1n-9
20:4n-6
Fatty Acid
C
intake
0.7
0.3
a
a
0.4
a
a
0.3
Pearson r
Pearson r
Adipose tissue
16:1n-7
0.2
0.5
0.2
0.6
0.1
a
a
0.1
b
0.1
16:0
16:1n-7
18:0
SFA
18:1n-9 20:4n-6
Fatty Acid
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
E4
0.2
CHO
P .001.
P < .01.
P < .05.
Sugar
Type of Intake
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
jamacardiology.com
Table 2. Adipose Tissue Fatty Acids and Cardiovascular Mortality in 853 Men With 251 Deaths
HR (95% CI)
Fatty Acid
P Value
Adjusted for
CVD Risk Factorsa
18:2n-6b
P Value
0.94 (0.83-1.07)
.37
0.92 (0.80-1.06)
.27
18:3n-3b
0.96 (0.83-1.11)
.59
0.99 (0.85-1.16)
.92
20:5n-3b
1.05 (0.93-1.19)
.44
1.05 (0.92-1.19)
.49
22:6n-3b
1.04 (0.9-1.19)
.53
1.01 (0.88-1.16)
.93
12:0c
0.87 (0.73-1.02)
.09
0.95 (0.78-1.15)
.58
14:0c
0.87 (0.73-1.03)
.10
0.95 (0.79-1.15)
.60
15:0c
0.96 (0.81-1.14)
.66
1.03 (0.86-1.23)
.74
16:0
1.08 (0.95-1.22)
.25
1.07 (0.94-1.22)
.28
16:1n-7
1.09 (0.96-1.24)
.19
1.03 (0.90-1.18)
.68
17:0c
0.93 (0.79-1.10)
.41
0.98 (0.82-1.16)
.79
18:0
0.87 (0.76-0.99)
.03
0.97 (0.84-1.12)
.71
18:1n-9
1.03 (0.89-1.18)
.69
1.04 (0.90-1.20)
.63
18:3n-6
0.96 (0.83-1.11)
.57
0.99 (0.85-1.15)
.86
20:3n-6
1.11 (0.96-1.28)
.15
0.94 (0.79-1.11)
.46
20:4n-6
1.25 (1.11-1.41)
<.001
1.13 (0.99-1.30)
.07
22:4n-6
1.26 (1.08-1.47)
.003
1.17 (0.98-1.40)
.08
22:5n-3
1.11 (0.97-1.27)
.11
1.04 (0.90-1.20)
.62
16:1n-7/16:0
1.07 (0.94-1.22)
.31
1.02 (0.89-1.16)
.81
18:1n-9/18:0
1.03 (0.89-1.18)
.69
1.04 (0.90-1.20)
.63
20:4n-6/18:2n-6
1.21 (1.08-1.36)
<.001
1.15 (1.01-1.31)
.04
Primary analyses.
Discussion
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,
jamacardiology.com
E5
Table 3. Adipose Tissue Fatty Acids and All-Cause Mortality In 853 Men With 605 Deaths
HR (95% CI)
Fatty Acid
P Value
Adjusted for
CVD Risk Factorsa
18:2n-6b
P Value
0.90 (0.82-0.98)
.01
0.90 (0.82-0.98)
.02
18:3n-3b
0.92 (0.84-1.01)
.08
0.96 (0.87-1.06)
.40
20:5n-3b
1.04 (0.95-1.12)
.40
1.03 (0.95-1.12)
.48
22:6n-3b
0.99 (0.91-1.08)
.90
0.97 (0.88-1.06)
.45
12:0c
0.89 (0.80-1.00)
.04
0.97 (0.85-1.10)
.64
14:0c
0.90 (0.81-1.00)
.06
0.97 (0.86-1.09)
.59
15:0c
0.91 (0.82-1.01)
.09
0.95 (0.85-1.07)
.42
16:0
1.05 (0.97-1.14)
.21
1.04 (0.96-1.13)
.31
16:1n-7
1.16 (1.07-1.26)
<.001
1.11 (1.02-1.21)
.02
17:0c
0.86 (0.77-0.97)
.01
0.89 (0.79-1.00)
.05
18:0
0.86 (0.80-0.94)
.001
0.92 (0.84-1.01)
.07
18:1n-9
1.08 (0.98-1.18)
.11
1.08 (0.98-1.18)
.12
18:3n-6
0.96 (0.87-1.05)
.38
0.97 (0.88-1.07)
.54
20:3n-6
1.07 (0.97-1.17)
.16
0.96 (0.86-1.06)
.41
20:4n-6
1.17 (1.08-1.27)
<.001
1.09 (0.99-1.19)
.07
22:4n-6
1.19 (1.07-1.31)
.001
1.10 (0.98-1.24)
.10
22:5n-3
1.07 (0.98-1.16)
.14
1.01 (0.92-1.11)
.80
16:1n-7/16:0
1.12 (1.04-1.22)
.005
1.08 (0.99-1.18)
.08
18:1n-9/18:0
1.08 (0.98-1.18)
.11
1.08 (0.98-1.18)
.12
20:4n-6/18:2n-6
1.19 (1.10-1.28)
<.001
1.13 (1.04-1.23)
.005
Primary analyses.
ably 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
jamacardiology.com
Hazard Ratio
0
5
10
15
20
25
30
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.
tality. 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 allcause 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 interac-
jamacardiology.com
E7
ARTICLE INFORMATION
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, rnlv, Risrus.
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: rnlv, Risrus.
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,
Nestl, 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.
REFERENCES
E8
Conclusions
jamacardiology.com
jamacardiology.com
E9