There were 126 participants who were randomized, of whom 118 completed the trial. For this secondary analysis, there were 113 plasma samples available at trial entry, 109 samples available post-supplementation (34–36 weeks) as well as 98 cord blood samples. The flow of participants and samples is shown in Figure 1.
The baseline characteristics of the three randomized groups are shown in Table 1. There were no significant differences in the baseline characteristics between the three randomized groups.
TABLE 1
Demographics
Variable
|
EPA-rich fish oil group N=36
|
DHA-rich fish oil group N=37
|
Soy oil placebo group N=41
|
Statistical Significance
|
Maternal age at enrollment
|
29.9+/-5.2
|
30.7+/-4.4
|
30.4+/-5.9
|
NS*
|
White race
|
32
|
28
|
34
|
NS‡
|
BMI at enrollment
|
28.5+/-6.3
|
28.5+/-6.7
|
27.7+/-8.2
|
NS*
|
Gestational age at enrollment (weeks)
|
15.9/2.7
|
16.9/2.3
|
16.0/2.3
|
NS*
|
Gravidity
|
2.42/1.36
|
2.55/1.22
|
2.41/2.02
|
NS‡
|
Parity
|
0.87/0.84
|
1.08/0.94
|
0.85/1.20
|
NS‡
|
NS = not significant *1-way ANOVA ‡ Fisher’s Exact test
Note: This table represents the subset of the total cohort with samples available for analysis. The total study demographics have been previously described.16
Our primary intent-to-treat analysis using repeated-measures modeling showed no significant difference in adiponectin, leptin, and the ALR in maternal plasma between the treatment groups and placebo group at baseline or after supplementation. (Table 2). These results were unchanged after square root transformation of the data.
TABLE 2
Adipokine values before and after supplementation according to treatment group
|
|
|
|
|
RM ANOVA P values
|
Parameter
|
Visit
|
EPA
|
DHA
|
Placebo
|
Group
|
Visit
|
Interaction
|
Adiponectin
ng/ml
|
1
|
24.15 (17.72,29.70)
|
26.15 (14.94,38.04)
|
25.29 (15.56,33.69)
|
0.83
|
<0.001
|
0.97
|
3
|
18.69 (13.79,22.05)
|
19.91 (12.95,27.90)
|
18.92 (14.20,25.71)
|
|
|
|
Leptin
ng/ml
|
1
|
27.93 (17.38, 35.77)
|
22.04 (11.46,34.44)
|
19.72 (15.04,28.34)
|
0.13
|
0.32
|
0.86
|
3
|
28.70 (19.69, 47.33)
|
24.38 (15.97,33.34)
|
23.91 (16.97,32.72)
|
|
|
|
A:L Ratio
|
1
|
0.83 (0.41,1.63)
|
1.44 (0.44,2.32)
|
1.17 (0.52,1.97)
|
0.60
|
<0.001
|
0.92
|
3
|
0.54 (0.34, 0.91)
|
0.77 (0.50,1.54)
|
0.75 (0.36,1.26)
|
|
|
|
Note. Cell formats are median (IQR). Reported adiponectin concentrations after 1:500 dilution.
Of interest, for the total cohort adiponectin significantly decreased between enrollment and 33–36 weeks gestation (p<0.001) while leptin significantly increased during the same time period (p = 0.008). The ALR significantly decreased during this same time period (p<0.001) indicating increasing insulin resistance with advancing pregnancy. The magnitude of decrease in the ALR was 31%.
Detailed comparison of change (delta) among the 3 treatment groups found no significant differences among the three groups of delta adiponectin, leptin, or the ALR. (Table 3).
Table 3.
Delta Adiponectin, Leptin, and ALR according to treatment group.
|
|
|
|
P values
|
|
Parameter
|
EPA
|
DHA
|
Placebo
|
Group
|
|
Δ Adiponectin
ng/ml
|
-5.35 (-30.0,6.43)
|
-5.17 (-19.92,16.44)
|
-6.66 (-41.10,15.27)
|
0.75
|
|
Δ Leptin
ng/ml
|
2.83 (-27.93, 35.03)
|
-0.05 (-40.93,17.60)
|
3.39 (-25.19,19.48)
|
0.45
|
|
Δ A:L Ratio
|
-0.47 (-5.47,0.37)
|
-0.51 (-4.16,0.40)
|
-0.68 (-6.11,0.61)
|
0.68
|
|
Note: P-values for an among group difference were computed using 1-way ANOVA. Reported adiponectin concentrations reflect 1:500 dilution.
Initiation of antidepressant medications during the trial was not significantly predictive of the ALR (P = 0.13).
We performed regression analyses to explore the relationships between measured serum DHA and EPA fraction at enrollment and after supplementation with measures of insulin sensitivity. Variables included in the model included BMI at enrollment, DHA fraction, EPA fraction, vitamin D concentration, and age at enrollment. After adjusting for BMI and maternal age, higher maternal serum DHA fraction was positively associated with increased ALR before (p = 0.01) and after (p = 0.04) supplementation. DHA fraction was positively associated with adiponectin after supplementation (p = 0.03); there was a non-significant trend towards association of DHA with adiponectin at study enrollment. There was no significant association of EPA with any measure of insulin sensitivity. Vitamin D concentration was also positively associated with adiponectin (p<0.05) at enrollment but was not significantly associated with adiponectin after supplementation.
Early pregnancy BMI was positively associated with maternal leptin levels at baseline and in late pregnancy (p< 0.001) and was inversely associated with the ALR (p< 0.001). ALR decreased significantly between the early and late pregnancy visit (p< 0.001). After adjusting for baseline weight, maternal weight gain between the early pregnancy visit and the late pregnancy visit was positively associated with plasma leptin (p<0.01) and negatively associated with the ALR (p<0.02)..
The effect of a one unit increase in each of the variables of interest on the ALR is shown in Table 4.
TABLE 4
Variables predictive of the change (delta) in the ALR
Variable
|
Estimate
|
95 % CI
|
Pregnancy Weight Gain (1 kg)
|
-0.04 (p<0.02)
|
[-0.07, -0.007]
|
Study Enrollment
BMI
|
-0.03 (p<0.0001)
|
[-0.04, -0.02]
|
After Supplementation
BMI
|
-0.02 (p<0.001)
|
[-0.03, -0.01]
|
Study Enrollment
DHA (1 unit)
|
0.2 (p<0.02)
|
[0.02, 0.17]
|
After Supplementation
DHA (1 unit)
|
0.1 (p<0.02)
|
[0.003, 0.112]
|
Note. P-values and Confidence Intervals (CI) were based on two-tailed student’s t distribution.
There were no differences in adiponectin, leptin, or the ALR in umbilical cord blood according to maternal treatment allocation group. (Table 5) Regression analysis was performed to evaluate the association of EPA fraction, DHA fraction, vitamin D concentration, birth weight, gestational age at delivery, and mode of delivery with adiponectin, leptin and the ALR in umbilical cord blood. There were no significant associations between the variables of interest on cord blood adiponectin or the ALR. However, birth weight and cord blood DHA fraction were significantly associated with serum leptin (p = 0.02, both). The association of birth weight with cord blood leptin persisted when controlling for DHA fraction, and the association of DHA fraction with cord blood leptin persisted when controlling for birth weight.
TABLE 5
Cord blood adipokine values according to treatment group
Parameter
|
EPA
|
DHA
|
Placebo
|
Group
|
Adiponectin
ng/ml
|
52.07 (38.99,70.98)
|
55.67 (38.81,68.34)
|
52.40 (43.42,65.70)
|
0.93
|
Leptin
ng/ml
|
9.25
(4.60,13.55)
|
7.98
(4.26,19.84)
|
7.10
(3.23,13.71)
|
0.56
|
A:L Ratio
|
5.85
(3.63,11.19)
|
6.21
(2.47,15.07)
|
8.69
(4.03,19.36)
|
0.27
|
Note. Cell formats are median (IQR).