Clinical Science (1992) 83, 97-101 (Printed in Great Britain)
97
Diet as a source of phospholipid esterified
9,l I-octadecadienoic acid in humans
Mary BRITTON, Christopher FONG*, David WICKENS* and John YUDKIN
Academic Unit of Diabetes and Endocrinology, and *Department of Chemical Pathology,
Whittington Hospital, London, U.K.
(Received 6 November 1991/14 February 1992; accepted 26 February 1992)
1. Diene-conjugated fatty acids are one of the products of free-radical attack upon lipids and therefore
have been used as markers of such attack. The major
diene-conjugated fatty acid in human tissue and
serum is an isomer of linoleic acid (9,12-octadecadienoic acid), namely 9’1 I-octadecadienoic acid. Diet
may be another source of this isomer, raising
questions as to its value as a free-radical marker. The
aim of this study was to determine the importance of
diet as a source of 9,ll-octadecadienoic acid in
phospholipid esterified fatty acids in human serum.
2. Foodstuffs rich in 9,ll-octadecadienoic acid were
identified. Fourteen subjects volunteered to alter their
diets, either increasing (‘high diet’) or decreasing
(‘low diet’) their intake of these foodstuffs for 3
weeks. Where subjects undertook both diets, a washout period of at least 3 weeks was allowed between
phases.
3. Seven-day diet histories were kept and scored with
respect to their content of 9’11-octadecadienoic acid.
The concentrations of 9,ll-octadecadienoic acid and
linoleic acid in serum phospholipids were measured by
h.p.1.c. with U.V. detection.
4. The percentage molar ratio of 9’11-octadecadienoic acid to linoleic acid was calculated. The
percentage molar ratio rose significantly on the ‘high
diet’ [1.3(0.4) versus 1.9(0.7), P= 0.01, mean (SD)]
and fell significantly on the ‘low diet’ [1.6(0.4) versus
l.l(O.4)’ P=0.004, means (SD)]. There was a significant correlation between the change in dietary intake
of 9,ll-octadecadienoic acid and the change in the
percentage molar ratio (r=0.829, P= 0.001).
5. The concentration of 9,ll-octadecadienoic acid in
serum phospholipids is influenced by diet. Its use as a
marker of free-radical activity is questionable and at
least in need of careful interpretation.
also been proposed that FR attack on low-density
lipoprotein (LDL) may initiate a chain of reactions
resulting in a modified, oxidized LDL more atherogenic than native LDL [CS].
Most methods available for the measurement of
FR species are unfortunately indirect and nonspecific [9, lo]. One approach used is to measure
concentrations of compounds thought to be products of FR attack upon endogenous molecules.
Diene-conjugated fatty acids (DCFAs) are one such
group of compounds, as they are one of the products of FR attack upon polyunsaturated fatty acids
(Fig. 1) [ll]. With respect to the aetiology of
atherosclerosis, these compounds are interesting not
only as possible markers of FR activity per se, but
in particular as evidence of FR attack on lipids, as
they are the product of the first step in lipid
Polyunsaturated
fatty acid
R\c/C,,c,C,,P
/ \
k
HI
Hydrogen abstraction
R\.,/C
I
Carbon-centred
radical
\‘,C,,P
,c.
0
I
Peroxyl radical 8
H
H
DCFA
0
INTRODUCTION
Interest has grown in the role of free-radical (FR)
species as mediators of tissue damage in many
disease processes, including diabetes [1-51. It has
b
Hydroperoxide
(also diene-conjugated) H
Fig. 1. Diene conjugate formation during FR attack on polyunsaturated fatty acids
Key words: diene-conjugated fatty acids, diet, free radicals.
Abbreviations: 18:2(9,1 I), 9,l I-octadecadienoic acid; 18:2(9,12), linoleic acid (9,12-octadecadienoicacid); DCFA, diene-conjugated fatty acid; FR, freeradical; LDL, lowdensity
lipoprotein; % MR, percentage molar ratio.
Correspondence: D r M. Britton, Academic Unit of Diabetes and Endocrinology, Whittington Hospital, Archway Road, London N I9 5NF, U.K.
D
98
M. Britton et al.
oxidation (Fig. 1) and possibly in the generation of
an atherogenic, modified LDL.
DCFAs can be measured spectrophotometrically.
The electronic arrangement in a diene-conjugated
system results in all compounds having such a
system demonstrating U.V.absorbance with maxima
in the range 23&235 nm. Total diene conjugation
can be measured using absorbance methods alone,
absorbance reflecting a contribution from all
DCFAs including diene-conjugated hydroperoxides
(Fig. 1). Thus such results could be only semiquantitative and given in arbitrary units. A combination of h.p.1.c. with U.V. detection allows individual specific diene-conjugated compounds to be
identified and measured in absolute concentrations
[12]. This has led to the identification of the major
DCFA in human tissues as 9,ll-octadecadienoic
acid, the 18:2(9,11) isomer of linoleic acid [9,12octadecadienoic acid 18:2(9,12)] [13]. Concentrations of 18:2(9,11) have been used as a measure
of FR activity. These concentrations are often
expressed as the percentage molar ratio (%MR) to
the parent compound 18:2(9,12) [12, 131 on the
assumption that its only source was FR attack on
this parent molecule.
More recently, other potential sources of DCFAs
have been recognized for humans [14, 151. Foodstuffs, mainly those of ruminant origin, have been
shown to be rich in 18:2(9,11) [16], but how well
these are absorbed is unclear. Fogerty et al. [17]
have shown foodstuffs to have a very wide range of
18.2(9,11) contents e.g. samples of lean beef contained between 6 and 43mg of 18.2(9,11)/100g of
total fat. This fatty acid was mainly distributed in
the fat component of the meat, beef fat containing
96&1310mg/100g. Butter contained between 720
and 910mg/100g. Fogerty et al. [17] also showed
that some foodstuffs of non-ruminant origin, e.g.
margarine, contained no 18,2(9,1l), but that some
contained a range of concentrations explicable in
terms of the animal food chain. Eggs contained up
to 31mg/100g of total fat, eggs from hens feeding
free range, with no animal fat source, having no
detectable 18,2(9,11). Thus controversy now exists as
to the importance of diet as a source of this and
other DCFAs.
We wished to test the hypothesis that diet is a
source of the 18:2(9,11) in serum phospholipids by
assessing the effect of altering the intake of foodstuffs high in 18:2(9,11) in a group of healthy
subjects.
EXPERIMENTAL
Table I. Foodstuffs rich in lkZ(9,Il).Their intake was increased during
the ’high diet’ phase and reduced during the ‘low diet’ phase.
I. Red meat: beef, lamb, pork, bacon, ham
2. Meat products: pies, sausages, pasties, pate, meat spread
3. Animal fats: butter, lard, cream
4. Dairy foods: dairy ice-cream
5. Cream soups and sauces
6. Full-fat dairy cheeses, milk, yoghurts
Methods
Two diets were devised using the work of Fogerty
et al. [17] to identify foodstuffs rich in 18:2(9,11).
The ‘high diet’ was rich in these foodstuffs, whereas
the ‘low diet’ was poor in such foodstuffs. Stringent
dietary instructions were not given, but subjects
were simply given a list of foodstuffs (Table 1) either
to avoid or to increase their intake of. Subjects
recorded a 7-day diet history on each phase of the
diet. These were scored allowing a score of 1 for a
typical portion of each item appearing on the
prescribed list of foodstuffs. Bccause previous work
had shown an enormous range of 18,2(9,11) concentration in food of identical origin [17], we did not
attempt to use any more than a semi-quantitative
score of its intake. Semi-skimmed milk was scored
as 1 and full-fat milk as 2 (per half pint). An average
daily score of number of items eaten was calculated.
All subjects were studied at baseline on their normal
diet and again after a period of 3 weeks on one of
the test diets. Where both diets were undertaken,
the order was randomized and a washout period of
at least 3 weeks was allowed between phases. Four
subjects undertook both dietary phases, four only
the ‘high diet’ and six only the ‘low diet’. Fasting
venous serum samples were collected and were
stored at -80°C while awaiting assay. All assays
were performed within 8 weeks of sample collection,
sample pairs being analysed in the same batch.
Serum phospholipids were subjected to enzymic
hydrolysis and DCFAs were determined by h.p.1.c.
with U.V. detection at 234 nm. Concentrations of
18:2(9,11) and 18:2(9,12) were determined and
expressed as a percentage molar ratio (XMR).
Statistical analysis
Statistical analysis was performed comparing concentrations at baseline with those at the end of
dietary phase using paired Student’s t-tests and the
correlation of variables by linear regression analysis.
Values are expressed as mean (SD).
Subjects
Fourteen healthy subjects (three males, 11
females; mean age 35.1 years, SD 8.7 years) took
part in the study. All were non-smokers.
RESULTS
The daily diet score of subjects on the ‘low diet’
was significantly lower than at baseline on their
99
Diet and dieneconjugated fatty acids
r
3.25 (a)
2.75 2.25’ -
g 1.75 -
x 1.25
-
0.75 0.25
-
1.5
k
2.75 -
~~
-3.5 -3.0 -2.5 -2.0 -1.5 -1.0
0.25
-0.9
1
I
O.OO
Before
r
O.O0
After
Before
3.251
-1.5
t
Fig. 4. Relationship between change in diet score and change in
%MR. r=0.829, P=O.Ool. 0.
‘High diet’, A,‘low diet’.
Fig.2. Effect of ‘high diet’ (a) and ‘low diet’ (b) on %MR of
18:2(9,11) to l8:2(9,12), Vertical bars represent mean (ID).
score and the change in %MR also correlated
significantly (r = 0.829, P =0.001) (Fig. 4).
0
2.75
A
After
0
DISCUSSION
0.25 -
0
I
I
I
I
I
I
I
normal diet [0.6(0.5) versus 2.6(0.7) items/day,
n = 10, mean (SD), P=O.O003]. On the ‘high diet’ it
was significantly higher [5.3(0.8) versus 2.9( 1.0)
items/day, n = 8, P = O.OOl].
The concentration of 18:2(9,11) increased from
12.1(3.7) to 18.8(7.4)pmol/l on the ‘high diet’
( P = 0.006) and decreased from 14.3(6.7) to
8.9(4.7) pmol/l on the ‘low diet’ ( P = 0.01). Correcting
for changes in 18:2(9,12) by expressing the value as
the %MR to 18:2(9,12), the %MR increased on the
‘high diet’ [1.3(0.4) versus 1.9(0.7); P=O.Ol] and
decreased on the ‘low diet’ [1.6(0.6) versus l.l(O.4);
P=0.004] (Fig. 2). Of the 18 pairs of results, 17
changed in the direction predicted by the hypothesis. Eleven subjects kept diet histories, allowing 25
pairs of diet scores and related %MR to be analysed. There was no correlation between baseline
diet score and baseline %MR. When all diet scores
and %MR were analysed together, there was highly
significant correlation between diet score and %MR
(r=0.636, P=O.OOl) (Fig. 3). The change in diet
It has been argued that DCFAs in human tissues
are derived from FR attack on polyunsaturated
fatty acids [ls]. This led to their use as markers of
FR activity in various disease states [l-51. The
recognition of the same chemical entities in animal
tissues [14] and the realization that they were
derived from enzymic processes raised the possibility
of alternative sources in humans, including diet
itself. The enzymic processes identified in animals
are both endogenous to the animal, e.g. rat liver
microsomes desaturate
11-octadecenoic acid
[18:l(ll)] to 18:2(9,11) [19], and exogenous, e.g.
bacteria residing in the bovine rumen bihydrogenate
18:2(9,12) to octadecanoic acid (18:O) via an
18:2(9,11) intermediate [15].
Similar processes may also exist in human tissue.
Local production and raised concentrations of
DCFAs in the human cervix as a result of bacterial
colonization and activity has already been reported
[20]. Several bacterial species involved in respiratory pathologies have been shown capable of producing 18:2(9,11) in vitro [21].
The aim of this study was to determine whether
changes in the dietary intake of 18.2(9,11) would be
reflected in the serum concentration of this entity.
We wished the test diets to be different for that
individual, but to remain realistic. To this end,
dietary instructions to subjects were given as simple
guidelines, rather than the more artificial recommendation of strictly measured quantities of defined
foodstuffs. It was felt this less rigid approach would
also improve compliance.
Our results support the hypothesis that the serum
concentration of 18:2(9,11) is influenced by dietary
intake. Despite deliberately, uncontrived test diets,
100
M. Britton e t al.
the percentage change in %MR in human serum on
the ‘high diet’ ranged from 10% to 88% and on the
‘low diet’ from 7% to 60%. This suggests that
normal dietary variation in a population would also
be sufficient to have a major influence on serum
concentration of 18:2(9,11).
Seventeen of the eighteen dietary interventions
produced changes in the %MR in the direction
predicted by the hypothesis. The one paradoxical
result arose in the subject with the lowest baseline
%MR (0.54). It may be that it is impossible to lower
the concentration beyond a certain level, a level
determined by factors other than diet.
Dietary compliance and 18:2(9,11) intake were
estimated using a simple diet score. Acknowledging
its inaccuracy, this method was adopted for the
following reasons. Unlike better recognized dietary
components, the 18:2(9,11) content of very few
foodstuffs is known; it does not appear in food
tables. In known sources its concentration varies
widely and will be greatly influenced by fat content
[17]. Therefore even if careful weighed records of
dietary intake were kept, calculation of 18:2(9,11)
content with any degree of accuracy would be
impossible. On the other hand, considering the
items in Table 1, although a typical portion of red
meat would weigh more than a typical portion of
pat6 or cheese, its fat content would be less, therefore typical portions of these items might be
assumed to be more equivalent in terms of
18:2(9,11) content than initial inspection would suggest. Also, the errors introduced by this assumption
of near equivalence are likely to be small relative to
the differences produced in the comparison between
the presence or absence of a typical portion of these
items in the diet, e.g. changing butter [720mg
18:2(9,11)/100g] for margarine (0 mg/100 g) will
produce such relatively large alterations in
18:2(9,11) intake as to swamp errors in equivalence
between portions of steak and pat& Moreover, any
inaccuracy of this method would be likely to disguise, rather than to exaggerate, any relationship
with plasma DCFAs ratios.
The lack of correlation between diet score and
%MR at baseline may be due to the inaccuracies of
the diet score, may be a statistical problem owing to
small numbers, or again may imply that other
factors contribute to %MR besides diet. FR activity
may be that other source, but the extent of the
influence of diet alone is so great as to confound the
situation completely. When all pairs of dietary
scores and %MR are considered there is significant
correlation (Fig. 3). The fact that the regression line
for diet score versus %MR does not pass through
zero also suggests a minimum concentration produced by sources other than those identified here.
There is also remarkably good correlation between
change in diet score and change in %MR (Fig. 4),
suggesting that our dietary assessment and scoring
system does identify the major contributors to
serum concentration despite its simplicity.
Some workers have found low levels of DCFA in
a group of young insulin-dependent diabetic
patients despite the theory that in diabetic patients
FR activity is increased and relates to complications
[22]. It is possible that these findings could be
explained by the subjects consuming diets low in
animals fats. Similar considerations may explain our
findings of lower levels of DCFA in a group of
Asian diabetic patients than in a Caucasian group,
despite the former being more at risk of cardiovascular complications 1231.
In conclusion, the use of DCFA as a marker of
FR activity when comparing groups or individuals
over a prolonged period is limited, as diet is
perhaps the major source of these compounds in
humans. We have not looked at how much day-today variation there is in DCFA levels in an individual. If this were not great, sudden major alterations might still be attributed to bursts of FR
activity in short-term longitudinal studies, e.g.
within hours of a significant event such as myocardial infarction [24] or major surgery [25].
REFERENCES
I. Collier, A., Wilson, R., Bradley, H., Thomson, ].A. & Small, M. Free radical
activity in type 2 diabetes. Diabetic Med. 1990; 7, 27-30.
2. Jennings, P.E., Jones, A.F., Florkowski, C.M., Lunec, J. & Barnett, A.H.
Increased diene conjugates in diabetic subjects with microangiopathy. Diabetic
Med. 1987; 4, 452-6.
3. Jennings, P.E. & Barnett. A.H. New approaches t o the problem of diabetic
microangiopathy. Diabetic Med. 1988; 5, I 11-17.
4. Lunec, J.,Halloran, S.P., White, A.G. & Dormandy, T.L. Free radical oxidation
products in serum and synovial fluid in rheumatoid arthritis. J. Rheumatol.
1981; 0, 233-45.
5. Lunec, J. Free radicals: their involvement in disease processes. Ann. Clin.
Biochem. 1990; 27, 173-82.
6. Steinberg, D., Parthasarathy, S., Carew, T.E., Khoo, J. & Witztum, J.L. Beyond
cholesterol: modifications of lowdensity lipoprotein that increase its
atherogenicity. N. Engl. J. Med. 1989; 120, 915-25.
7. Bruckdorfer, K.R. Free radicals, lipid peroxidation and atherosclerosis.
Curr. Opin. Lipidol. 1990; I, 529-35.
8. Bedwell, S. & Jessup. W. Effects of oxygenxentred free radicals on low
density lipoprotein structure and metabolism. Biochem. SOC. Trans. 1987; 15,
259-60.
9. Halliwell, B. & Grootveld, M. The measurement of free radical reactions in
humans. FEBS Lett. 1987; 213, 9-14.
10. Knight, J.A., Pleper, R.K. & McClellan, L. Specificity of the thiobarbituric acid
reaction. Its use in studies of lipid peroxidation. Clin. Chem. 1988; 34,
24334.
I I. Halliwell, B. & Gutteridge, J. Free radicals in biology and medicine. Oxford:
Clarendon Press/Oxford University Press, 1985: 160-70.
12. Iversen, S.A., Cawood, P. & Dormandy, T.L. A method for the measurement
of diene conjugated derivative of linoleic acid. l8:2(9,1 I), in serum
phospholipid, and possible origins. Ann. Clin. Biohem. 1985; 22, 137-40.
13. Iversen, S.A., Cawood, P., Madigan, M.]., Lawson, A.M. & Dormandy, T.L.
Identification of a diene conjugated component of human lipid as
octadeca-9,1 Idienoic acid. FEBS Lett. 1984; 171, 320-4.
14. Ackman, R.G., Eaton, C.A., Sipos, J.C. & Crewe, N.F. Origin of cis-9,trans-ll,
and trans-9,trans-l I-octadecadienoic acids in the depot fats of primates fed a
diet rich in lard and corn oil and its implications for the human diet. Can.
Inst. Food Sci. Technol. J. 1981; 14, 102-7.
15. Hughes, P.E., Hunter, W.]. & Tove, S.B. Biohydrogenation of unsaturated fatty
Diet and diene-conjugated fatty acids
acids. J. Biol. Chem. 1982; 257, 3643-9.
16. Parodi, P.W. Conjugated octadecadienoic acids of milk fat. J. Dairy Sci. 1977;
60, 1550-3.
17. Fogerty, A.C., Ford, G.L. & Svoronos, D. Octadeca-9,1 ldienoic acid in
foodstuffs and in the lipids of human blood and breast milk. Nutr. Rep. Int.
1988; 38, 937-44.
18. Dormandy, T.L. & Wickens, D.G. The experimental and clinical pathology of
diene conjugation. Chem. Phys. Lipids 1987; 45, 353-64.
19. Holman, R.T. & Mafouz, M.M. Cis and trons octadecadienoic acids as
precursors of polyunsaturated acids. Prog. Lipid Res. 1981; 20, 151-6.
20. Fairbank, J.,Hollingworth, A., Griffin, J. et al. Octadeca-9.1 Idienoic acid in
cervical intraepithelial neoplasia: a colposcopic study. Clin. Chim. Acta 1989;
186,534.
21. Jack, C.I.A., Jackson, M.J., Ridgeway, E. & Hind, C.R.K. Octadeca-9,ll dienoic
22.
23.
24.
25.
I01
acid-a measurement of free radical activity o r a marker of infection in the
lung [Abstract]. Clin. Sci. 1991; 81 (Suppl. 25), 62P.
Collier, A., Jackson, M., Dawkes, R.M., Bell, D. & Clarke, B.F. Reduced free
radical activity detected by decreased diene conjugates in insulindependent
diabetic patients. Diabetic Med. 1988; 5, 747-9.
Britton, M.E., Hendra, T.J., Wickens. D. & Yudkin, J.S. Free radical activity
does not explain the increased atherosclerotic risk in diabetic o r Asian
subjects [Abstract]. Diabetic Med. 1989; 6, P67.
Hendra, T.]., Wickens, D.G., Dorrnandy, T.L. & Yudkin, J.S., Platelet function
and conjugated diene concentrations in diabetic and nondiabetic survivors of
acute myocardial infarction. Cardiovasc. Res. 1991; 25, 676-83.
Davies, S.W., Underwood, S.M., Wickens, D.G., Feneck, R.O., Dormandy, T.L.
& Walesby, R.K. Systemic patterns of free radical generation during coronary
bypass surgery. Br. Heart J. 1990; 64, 23640.