The SA Journal Diabetes & Vascular Disease Volume 16 No 2 (November 2019)

26 / 40

RESEARCH ARTICLE SA JOURNAL OF DIABETES & VASCULAR DISEASE
68 VOLUME 16 NUMBER 2 • NOVEMBER 2019
Among dietary FAs, the first extracted pattern presented with
high positive loadings of saturated FAs, MUFAs, α-linolenic acid
(C18:3n-3) and n-6 FAs, and therefore was named the ‘non-
marine’ FA pattern. The second pattern was named the ‘marine’ FA
pattern because it was characterised by high positive loadings of
eicosapentaenoic acid (C20:5n-3) and C22:6n-3.
The six plasma phospholipid FA patterns are discussed in the
order in which they were derived. The first pattern presented with
positive loadings of LC-SFAs, C18:0, C20:0, C22:0 and C24:0 and
very high negative loadings of C16:1n-7, C18:1n-7 and C18:1n-
9; we named it the ‘high-Satfat’ pattern. The second pattern was
named ‘n-3 VLC-PUFA’ and presented with high positive loadings
of docosapentaenoic acid (C22:5n-3), C22:6 n-3 and C20:5n-3, as
well as C20:4n-6. The third pattern presented the highest positive
loadings of C18:2n-6 and eicosadienoic acid (C20:2n-6) and was
named accordingly as the ‘high-LA’ pattern. The fourth pattern was
named ‘n-6 VLC-PUFA’ since it was characterised with high positive
loadings of adrenic acid (C22:4n-6), C22:2-n6 and C20:3n-6. The
fifth pattern extracted was named the ‘n-9 LC-MUFA’ pattern and
presented with positive loadings of C24:1n-9 and gondoic acid
(C20:1n-9). The sixth and last pattern had a positive loading of one
FA, i.e. C18:3n-3, and we named it ‘ n-3 EFA’ pattern.
Dietary FA patterns were weakly associated with measured
outcomes (Table 4). The non-marine FA pattern showed marginal
positive associations with WC in the crude model and the
association remained marginal after adjusting for age and gender
(β = 0.06, 95% CI = –0.01–0.13, p = 0.09). The association was
lost after adjustment for lifestyle variables and energy intake. On
the other hand, we did not find any associations with the marine
FA pattern (Table 4). Neither pattern revealed any association with
BMI, WHtR or the MetS. Further adjustment to the regressions for
total fat, fibre, carbohydrates and added sugar did not result in
any significant associations. The variables in the adjusted models
explained 0.02 to 27% of the variation in measures of adiposity
and 0.4 to 20% of the variation in the MetS.
Plasma phospholipid FA patterns resulted in stronger associations
with measures of adiposity and the MetS (Table 5). The high-Satfat
and n-3 VLC-PUFA patterns were positively associated with all
measures of adiposity and the MetS. The associations remained
significant in the fully adjusted model. The omega-6 VLC-PUFA
pattern showed marginal and positive associations with WC and
WHtR in the crude model, but associations were lost after further
adjustments. This pattern also showed higher odds for having the
MetS and remained significantly associated in the fully adjusted
model (odds ratio, OR = 1.25, 95% CI = 1.02–1.54, p = 0.03).
The n-9 LC-MUFA pattern was inversely associated with WC and
WHtR in the crude model as well as after adjustment for age and
gender. The associations were, however, lost after adjustments for
lifestyle variables and energy intake. This pattern also showed lower
odds for having the MetS and remained significantly associated in
the fully adjusted model (OR = 0.61, 95% CI = 0.50–0.75, p ≤
0.0001).
The omega-3 EFA pattern showed an inverse association with
BMI, WC and WHtR, but in the fully adjusted model marginal
significance remained for BMI only. This pattern also showed lower
odds for having the MetS and remained significantly associated in
the fully adjusted model (OR = 0.81, 95% CI = 0.66–0.99, p =
0.04). The variables in all the adjusted models explained 14 to 34%
of the variation in measures of adiposity, and 18 to 31% of the
variation in the MetS.
We further adjusted all regression models for use of
contraceptives and intakes of total fat, fibre, carbohydrates, and
energy from added sugar in association with plasma phospholipid
FAs. Additional adjustment for these variables did not result in
different associations with anthropometric indices. The association
between high-LA pattern and the MetS remained marginally
significant after adjusting for additional variables, whereas the
associations with the n-6 VLC-PUFA and n-3 EFA patterns were lost.
Discussion
The results of this study add new information about identified FA
patterns both in diet and plasma phospholipids among a selected
group of black South Africans from the North West Province. We
identified for the first time two dietary FA patterns and six plasma
phospholipid FA patterns (Table 3) by means of factor analysis
in this group of black adults. The dietary non-marine FA pattern
showed a weak positive association with WC, whereas the marine
pattern did not show any associations with outcomes measured.
On the other hand, two plasma phospholipid FA patterns
(high-Satfat and n-3 VLC-PUFA) were positively associated with
all measures of adiposity and the MetS. The omega-6 VLC-PUFA
pattern showed a positive association with the MetS, but not
with measures of adiposity. The n-9 LC-MUFA and the n-3 EFA
patterns showed an inverse association with the MetS in fully
adjusted models and tended to be negatively associated with some
measures of adiposity. The high-LA pattern was neither associated
with measures of adiposity nor the MetS. Our findings indicate
that dietary FA patterns were weakly associated, whereas plasma
phospholipid FA patterns were more strongly associated with
measures of adiposity and the MetS.
Previous studies have reported FA patterns, derived fromdifferent
components of blood and tissue in association with obesity 29 and
the MetS,22,30 but not with dietary patterns. These patterns were
generated by varying numbers of FAs ranging from nine to 34
FAs,22,29,30 and some included estimated desaturase activities, 30 by
means of use of factor29,30 and cluster 22 analysis. Consequently,
these derived patterns differed from that obtained in our study.
A dietary pattern, consisting of SFAs, PUFAs, MUFAs and
other nutrients, was not associated with obesity among Iranian
adults.45 On the contrary, a multiracial study in the USA reported
a positive association of intakes of total fat, total saturated fat,
LC-SFAs, myristic acid (C14:0), C16:0 and C18:0, and MUFAs with
BMI.46 Furthermore, a study investigating the association of dietary
patterns with the MetS concluded that a pattern high in meat
products was associated with a higher prevalence of the MetS.47
In our study, the dietary non-marine FA pattern showed marginal
and positive associations with WC, but not with other measures
of adiposity or the MetS. The non-marine FA pattern had positive
loadings of FAs from SFAs, MUFAs and PUFAs, specifically from
two SFAs (C16:0 and C18:0), two MUFAs (C16:1n-7, C18:1n-9)
and two PUFAs (C18:2n-6 and C18:3 n-3). The dietary marine FA
pattern showed no association with outcomes measured.
Our results are in agreement with a study in the USA that also
found no associations of n-3 LC-PUFAs with BMI due to low intakes
of these FAs in their participants.46 In our study and the study in
the USA, lower intakes of n-3 PUFA compared to the FAO/WHO
recommendation of 0.25–2 g/day were found.48 Under-reporting
of dietary intake may significantly influence nutrient pattern
investigation and association with disease, 49 however, in the PURE