Skip to main content
NCBI home page
UKPMC Funders Author Manuscripts logoLink to UKPMC Funders Author Manuscripts
. Author manuscript; available in PMC: 2018 Oct 1.
Published in final edited form as: Am J Clin Nutr. 2018 Jul 1;108(1):92–100. doi: 10.1093/ajcn/nqy079

Long-chain polyunsaturated fatty acids, gestation duration and birth size: a Mendelian randomization study using FADS variants

Jonathan Y Bernard 1, Hong Pan 1, Izzuddin M Aris 1, Margarita Moreno-Betancur 2,3, Shu-E Soh 1,4, Fabian Yap 5, Kok Hian Tan 6,7, Lynette P Shek 1,4,8, Yap-Seng Chong 1,9, Peter D Gluckman 1,10, Philip C Calder 11,12, Keith M Godfrey 11,12,13, Mary Foong-Fong Chong 1,14,15, Michael S Kramer 9,16, Neerja Karnani 1,17, Yung Seng Lee 1,4,8
PMCID: PMC6038907  EMSID: EMS78520  PMID: 29878044

Abstract

Background

In randomized trials, supplementation of omega-3 (ω3) long-chain polyunsaturated fatty acids (LC-PUFA) during pregnancy has resulted in increased size at birth, attributable to longer gestation.

Objective

We examined this finding using a Mendelian randomization approach utilizing fatty acid desaturase (FADS) gene variants affecting LC-PUFA metabolism.

Design

As part of a tri-ethnic mother-offspring cohort in Singapore, 35 genetic variants in FADS1, FADS2 and FADS3 were genotyped in 898 mothers and 1,103 offspring. Maternal plasma ω3 and ω6 PUFA levels at 26-28 weeks gestation were measured. Gestation duration was derived from a dating ultrasound scan in early pregnancy and birth date. Birth length and weight were measured. Eight FADS variants were selected through tagging-SNP approach and examined in association with PUFA levels, gestation duration among spontaneous labors and birth size, using ethnicity-adjusted linear regressions and survival models that accounted for the competing risks of induced labor and pre-labor caesarean section.

Results

Maternal FADS1 variant rs174546, tagging for 8 other variants located on FADS1 and FADS2, was strongly related to plasma ω6 but not ω3 LC-PUFA concentrations. Offspring and maternal FADS3 variants were associated with gestation duration among women who had spontaneous labor: each copy of rs174450 minor allele C was associated with a shorter gestation by 2.2 (95% CI: 0.9, 3.4) and 1.9 (0.7, 3.0) days for maternal and offspring variants, respectively. In survival models, rs174450 minor allele homozygotes had reduced time to delivery after spontaneous labor compared with major allele homozygotes (hazard ratio (95% CI): 1.51 (1.18, 1.95) and 1.51 (1.20, 1.89) for maternal and offspring, respectively).

Conclusions

Using a Mendelian randomization approach, we observed associations between FADS variants and gestation duration. This suggests potential role of LC-PUFAs in gestation duration.

Keywords: pregnancy, birth size, fatty acids, cohort studies, Mendelian randomization

Introduction

The omega-3 (ω3) and omega-6 (ω6) long-chain polyunsaturated fatty acids (LC-PUFAs), such as docosahexaenoic acid (DHA, 22:6ω3) and arachidonic acid (AA, 20:4ω6) are required by the fetus for normal development of the brain, retina and central nervous system (1–3). Experimental and clinical evidence shows that the placenta, a tissue of fetal origin, transfers maternal LC-PUFAs to the fetal circulation (4).

The ω3 and ω6 LC-PUFAs can be provided through the diet or synthetized from their respective precursors, alpha-linolenic (ALA, 18:3ω3) and linoleic (LA, 18:2ω6) acids. ALA and LA cannot be synthesized in mammals and are required in the diet (5). LC-PUFA synthesis from their precursors involves desaturation and elongation (Supplemental Figure 1), which are known to be inefficient in humans, especially for ω3 PUFAs (6). LC PUFA synthesis is more efficient in women than in men, and may be enhanced during pregnancy (7–9). It is controlled by the rate-limiting enzymes delta-5 and delta-6 desaturases, which are encoded by the FADS1 and FADS2 genes, respectively (10, 11). Located on the same gene cluster, FADS3 is presumed to have a role in PUFA metabolism; its encoded protein and exact function remain unknown (11, 12). Single nucleotide polymorphisms (SNP) within the FADS cluster are related to plasma, erythrocyte and breast milk PUFA content and to child health outcomes (11, 13–16).

Meta-analyses of randomized controlled trials suggest that supplementation of ω3 LC-PUFAs or fish oil (rich in ω3 LC-PUFAs) during pregnancy leads to a reduced risk of early preterm delivery and small increases in birth weight and length attributable to a longer gestation (17–19). Most, but not all, observational studies reported similar findings with maternal fish consumption as exposure (20–22). Yet, published RCTs and observational studies have limitations preventing from definitive inferences about causality (23, 24). The published RCTs were carried out in Western populations only, and less is known in other populations where both genetic background and LC-PUFA intake differ. Observational studies, mostly carried out in Western populations as well, are subject to confounding and reverse causality. Varying study designs and approaches may help overcome inherent pitfalls and strengthen the evidence base (25). Mendelian randomization (MR) has been proposed as a method to remove the risk of reverse causality and confounding, by using genetic variants as instrumental variables for the exposure of interest (26).

Using data from the multiethnic mother-offspring cohort Growing Up in Singapore Towards healthy Outcomes (GUSTO), we examined whether maternal and offspring FADS gene variants, used as instrumental variables for maternal and antenatal LC-PUFA status, are related to gestation duration and size at birth.

Subjects and Methods

Study population

Between June 2009 and September 2010, the GUSTO study recruited pregnant women of Chinese, Malay and Indian ethnicities and attending before the 15th week of pregnancy their first ultrasound scan at two public maternity units in Singapore. Women were ineligible for the study if they had non-homogeneous ethnic background (up to the four grandparents of the offspring), had no intention of delivering in the study centers or not staying in Singapore for the 5 years following delivery. The recruitment protocol has been described in detail (27). All participants signed informed written consent at enrolment. The study received ethical approval from the National Healthcare Group Domain Specific Review Board and the SingHealth Centralised Institutional Review Board. From the 1237 pregnant women recruited, 1171 singleton newborns were included. Twins were excluded from the current analysis (Supplemental Figure 2). Information on gestation duration, whether the labor was spontaneous or induced, and offspring sex were abstracted from birth records.

Maternal PUFA levels

Fasting blood samples were collected at 26-28 weeks’ gestation. Blood lipid extraction was undertaken with chloroform/methanol (Fisher Scientific, Hampton, NH, USA). Plasma phosphatidylcholine (PC) was separated by solid-phase extraction. Fatty acid methyl esters were prepared by reaction with methanolic sulphuric acid and then separated by gas chromatography (BPX-70 column mounted on a Hewlett-Packard HP6890) and detected by flame ionization before quantification as μg/mL of plasma. Inter- and intra-assay variation coefficients for all fatty acids identified in plasma PC were lower than 6% and 3%, respectively. Eleven PUFAs were identified (Supplemental Figure 1) and expressed as % of total fatty acids. Additional details have been published previously (28).

Genotyping and selection of FADS variants

Maternal and offspring DNA were extracted from maternal blood and umbilical cord, respectively. Samples were genotyped using Illumina Omniexpress + exome arrays (Illumina Inc., San Diego, CA, USA). DNA hybridization arrays and subsequent scanning were performed by Expression Analysis Inc. (Morrisville, NC, USA). Data were processed in GenomeStudio Genotyping Module v1.0 (Illumina Inc.).

We focused on FADS genes since they were reported being the most strongly related to LC-PUFA status, including in Asian populations (11, 13–16, 29, 30). Maternal and offspring genotypes of 35 variants from the FADS genes region (chromosome 11, 11q12-13.1) were available on our arrays. All were located on introns between positions 61569830 and 61656117. Data quality control procedures excluded the following: i) rate of missingness >5% in the whole sample (no variants), ii) monomorphism in the overall cohort (13 variants), iii) minor allele frequency <5% by ethnicity (5 variants), or iv) p-value for Hardy-Weinberg Equilibrium test by ethnicity <10-3 (no variants). Seventeen variants were retained after this quality control step (Supplemental Table 1). Linkage disequilibrium plots in mothers and offspring are shown in Supplemental Figures 3 and 4, respectively. Because high linkage disequilibrium was observed between variants, we then used a tagging-SNP approach with Haploview software v4.2 (Broad Institute of MIT and University of Harvard, Boston, MA, USA). Briefly, this method selects variants that are representative of regions in high linkage disequilibrium, and allows examination of a smaller number of variants that are less likely to be inherited together. Pairwise tagging-SNP selection was performed based on r2 using a threshold of ≥0.80. The 17 variants were tagged by 8 variants as follows: rs174546 on FADS1, rs2727270, rs174593, rs498793 and rs17156506 on FADS2, and rs174450, rs1000778 and rs174455 on FADS3 (Supplemental Table 1).

Gestation duration and anthropometric measurements at birth

Gestational age and date of conception were estimated by ultrasonography when women attended their first dating scan during the first trimester of pregnancy (mean gestational age ± SD: 11.7 ± 1.7 weeks, range: 6.6-17.0). Gestation duration was derived from the estimated dates of conception and of delivery, and was reported in days. The offspring’s anthropometric measures were taken in duplicate by trained hospital staff within a few hours of birth and then averaged. Weight was measured to the nearest 1 g using a calibrated mobile digital scale (Seca 334, Seca GmBH & Co Kg, Hamburg, Germany). Length was measured in the recumbent position to the nearest 1 mm using a mobile infant mat (Seca 210). All research staff were unaware from the participants’ FADS genotype and maternal mid-pregnancy PUFA status. Gestational age-specific weight and length z-scores at birth for Singaporeans were derived from a previous reference based on our cohort (31).

Mendelian randomization (MR)

This study relies on a MR design, a design proposed to remove reverse causality and confounding (26). Figure 1 shows the directed acyclic graph (DAG) (24) that illustrates the MR design as applied to our study to estimate the causal effect of LC-PUFA status (exposure E) on gestation duration (outcome O) using FADS genetic variants (G) as instrumental variables for E. If G causes E (at least partly), has no effect on O other than through its effect on E, and is not associated with any E→O confounder U, then the relationship E→O can be estimated from the relationship G→O. Although maternal LC-PUFA intake (V) is a major determinant of E, it is not a confounder of the E→O and G→O relationships because it is unrelated to FADS variants and must not be controlled for (FADS genetic variants do not affect dietary LC-PUFA intake, and vice-versa). Potential confounding in estimating G→O may, however, arise from ethnicity (W), which affects G, E, O, and also V, and therefore should be controlled for.

Figure 1.

Figure 1

Open in a new tab

Directed acyclic graph (DAG) representing the Mendelian randomization approach to estimate the causal effect of LC-PUFA status (E) on gestation duration (O) using FADS genetic variants (G) as instrumental variables. U, V and W denote potential unmeasured confounders, maternal dietary PUFA intake and ethnicity, respectively. Unmeasured factors (U) confound the relationship E→O, but not G→O. Dietary PUFA intake (V) is related to O only through E and therefore cannot confound the E→O or G→O relationships. Ethnicity (W) affects LC-PUFA status (E) through FADS variants (G) and dietary PUFA intake (V) and is therefore a confounder of both the E→O and G→O relationships.

Statistical analyses

As a preliminary analysis, we assessed the validity of using maternal FADS variants as instrumental variables within our population of Asian pregnant women by examining the associations of maternal FADS variants with maternal PUFA levels and ratios. This was carried out by linear regression: variants were first introduced into the models as categorical variables; since a codominant effect was observed in heterozygotes, we then used an additive genetic model by introducing variants as ordinal variables. We therefore report regression coefficients for each additional copy of the minor allele (major allele homozygous as reference). PUFA variables were standardized (mean=0, SD=1) to facilitate comparisons of regression coefficients across tests.

Associations of maternal and offspring FADS variants with gestation duration were examined by linear regression among the mother-offspring pairs who experienced spontaneous labor (excluding births after labor induction and pre-labor cesarean delivery). In a complementary approach, we examined the associations between FADS variants and gestation duration by using competing risks survival models for time to birth. In these models, spontaneous delivery after spontaneous labor was considered as the event of interest, while induced labor and pre-labor cesarean delivery were competing events. Indeed, FADS gene variants were associated with type of labor (Supplemental Table 2); not accounting for competing events in survival models may result in bias (32). We therefore used both cause-specific proportional hazards models and the Fine–Gray proportional subdistribution hazards models, yielding cause-specific hazard ratios (csHRs) and subdistribution hazard ratios (sdHRs), respectively (32). We reported the HRs for both the event of interest (spontaneous labor) and the competing events (induced labor and pre-labor caesarian delivery) (33). As a prerequisite, we assessed the proportional hazards assumption by plotting the Schoenfeld residuals (cause-specific hazard model) and the cumulative incidence function (subdistribution hazard model). Since potential departures from that assumption were observed, we performed sensitivity analyses by allowing a time-varying effect by adding the time*FADS interaction into the models. We plotted the cumulative incidence function when associations of FADS variants with gestation duration were observed. Finally, the associations of FADS variants with gestational age-adjusted birth size (length and weight) among offspring with birth after spontaneous labor were analyzed by linear regression.

All analyses were carried out both in the full study sample (with adjustment for ethnicity) and after stratification by ethnicity. Previous publication based on genome-wide data in GUSTO reported excellent agreement between self-reported ethnicity and genetic ancestry clustering (34). All statistical analyses were performed with SAS 9.4 (SAS Institute Inc, Cary, NC, USA).

Results

The flow diagram for study participation is shown in Supplemental Figure 1. Maternal (n=897) and offspring (n=1100) genotype frequencies for the 17 variants that passed quality control are shown in Table 1. Important frequency differences in FADS1 and FADS2 variants were observed by ethnicity; major alleles in Chinese and Malay participants were opposite to those in Indian participants (p-value χ2 test=10-100 and 10-18 for rs174546 and rs174593, respectively). In Indian participants, the frequency of minor allele homozygotes was generally lower than 1%. FADS3 variant frequencies were more balanced across ethnicity but remained significantly different (p-value=10-10 for rs174450).

Table 1.

Maternal and offspring FADS genotype frequencies by ethnicity in the GUSTO cohort.

SNPs Position (kb) Alleles (M/m)1 Maternal allele frequency by ethnicity (MM / Mm / mm) Offspring allele frequency by ethnicity (MM / Mm / mm)
Chinese
n=516		 Malay
n=231		 Indian
n=151		 Chinese
n=628		 Malay
n=277		 Indian
n=198
FADS1
   rs174546 61569830 A/G 0.37 / 0.46 / 0.17 0.56 / 0.38 / 0.05 0.01 / 0.23 / 0.77 0.33 / 0.51 / 0.16 0.57 / 0.38 / 0.05 0.01 / 0.19 / 0.81
   rs174547 61570783 G/A 0.37 / 0.46 / 0.17 0.56 / 0.38 / 0.05 0.01 / 0.23 / 0.77 0.33 / 0.51 / 0.16 0.57 / 0.38 / 0.05 0.01 / 0.19 / 0.81
   rs174548 61571348 C/G 0.35 / 0.47 / 0.18 0.51 / 0.40 / 0.07 0.01 / 0.22 / 0.77 0.32 / 0.52 / 0.16 0.52 / 0.41 / 0.07 0.00 / 0.17 / 0.83
   rs174550 61571478 G/A 0.37 / 0.46 / 0.17 0.56 / 0.37 / 0.05 0.01 / 0.23 / 0.77 0.33 / 0.51 / 0.16 0.57 / 0.38 / 0.05 0.01 / 0.19 / 0.81
FADS2
   rs174570 61597212 A/G 0.36 / 0.46 / 0.18 0.55 / 0.37 / 0.06 0.00 / 0.15 / 0.85 0.32 / 0.52 / 0.16 0.53 / 0.40 / 0.07 0.01 / 0.13 / 0.86
   rs1535 61597972 G/A 0.37 / 0.46 / 0.17 0.57 / 0.37 / 0.05 0.01 / 0.21 / 0.78 0.33 / 0.51 / 0.16 0.57 / 0.38 / 0.05 0.01 / 0.19 / 0.81
   rs2727270 61603237 G/A 0.29 / 0.50 / 0.21 0.26 / 0.55 / 0.14 0.87 / 0.13 / 0.00 0.29 / 0.54 / 0.18 0.29 / 0.48 / 0.24 0.91 / 0.09 / 0.00
   rs174576 61603510 A/C 0.37 / 0.46 / 0.17 0.58 / 0.35 / 0.05 0.01 / 0.23 / 0.76 0.33 / 0.51 / 0.16 0.57 / 0.38 / 0.04 0.01 / 0.19 / 0.80
   rs174577 61604814 A/C 0.37 / 0.46 / 0.17 0.58 / 0.36 / 0.05 0.01 / 0.23 / 0.77 0.33 / 0.51 / 0.16 0.57 / 0.38 / 0.05 0.01 / 0.19 / 0.80
   rs174583 61609750 A/G 0.37 / 0.45 / 0.17 0.58 / 0.35 / 0.05 0.01 / 0.28 / 0.72 0.34 / 0.51 / 0.16 0.58 / 0.38 / 0.05 0.02 / 0.19 / 0.79
   rs2851682 61616012 A/G 0.31 / 0.49 / 0.20 0.26 / 0.56 / 0.13 0.88 / 0.12 / 0.00 0.29 / 0.54 / 0.17 0.29 / 0.48 / 0.23 0.91 / 0.09 / 0.00
   rs174593 61618831 A/G 0.73 / 0.25 / 0.03 0.45 / 0.46 / 0.06 0.77 / 0.22 / 0.01 0.72 / 0.26 / 0.02 0.48 / 0.41 / 0.11 0.79 / 0.20 / 0.01
   rs498793 61624705 G/A 0.83 / 0.16 / 0.01 0.66 / 0.30 / 0.03 0.40 / 0.49 / 0.11 0.84 / 0.16 / 0.01 0.69 / 0.27 / 0.04 0.40 / 0.46 / 0.13
   rs174611 61627881 A/G 0.97 / 0.03 / 0.00 0.88 / 0.11 / 0.01 0.87 / 0.13 / 0.00 0.98 / 0.02 / 0.00 0.85 / 0.14 / 0.01 0.89 / 0.11 / 0.00
   rs174618 61629322 A/G 0.96 / 0.04 / 0.00 0.82 / 0.16 / 0.01 0.52 / 0.42 / 0.06 0.96 / 0.04 / 0.00 0.78 / 0.20 / 0.03 0.57 / 0.39 / 0.05
   rs17156506 61632913 G/A 0.85 / 0.13 / 0.01 0.77 / 0.21 / 0.01 0.95 / 0.05 / 0.00 0.85 / 0.15 / 0.00 0.78 / 0.21 / 0.01 0.96 / 0.04 / 0.00
FADS3
   rs174450 61641542 A/C 0.34 / 0.49 / 0.16 0.17 / 0.47 / 0.32 0.27 / 0.53 / 0.20 0.35 / 0.50 / 0.15 0.21 / 0.45 / 0.35 0.28 / 0.46 / 0.25
   rs1000778 61655305 G/A 0.49 / 0.42 / 0.09 0.43 / 0.43 / 0.10 0.31 / 0.48 / 0.21 0.49 / 0.43 / 0.08 0.46 / 0.39 / 0.15 0.30 / 0.52 / 0.18
   rs174455 61656117 A/G 0.37 / 0.47 / 0.16 0.20 / 0.47 / 0.28 0.26 / 0.50 / 0.24 0.35 / 0.51 / 0.14 0.22 / 0.44 / 0.33 0.27 / 0.50 / 0.23
Open in a new tab
1

Major (M) and minor (m) alleles were defined based on allele frequencies in the whole study.

Tagged SNPs are in bold.

Most maternal plasma LC-PUFA levels differed by ethnicity (Supplemental Table 3). This was particularly striking for ω6 LC-PUFA levels, which were higher in Indian women, and for ω3 LC-PUFA levels, which were higher in Chinese women. The preliminary analysis comparing maternal plasma PUFA levels by maternal variants confirmed the validity of using FADS variants as instrumental variables (Supplemental Table 4). Strong associations (p-values <10-7) were seen between plasma ω6 PUFA levels and maternal FADS1 rs174546 and FADS2 rs2727270 (both being in negative linkage disequilibrium). Minor allele carriers had higher and lower levels of plasma ω6 LC-PUFA levels, respectively. The associations were less strong for the variants located on FADS3; minor allele carriers of rs174450, rs1000778 and rs174455 had lower plasma ω6 LC-PUFA levels (p-values between 10-3 and 10-7). In contrast, no association between FADS variants and ω3 PUFA levels reached a significance level <10-3. These associations were, overall, similar by ethnicity (Supplemental Table 5).

Among mother-offspring pairs who experienced birth by spontaneous labor, all three maternal FADS3 variants were associated with gestation duration (Table 2). For rs174450, each additional copy of the minor allele C (associated with lower LC-PUFA levels) was associated with a shorter gestation by 2.2 (95% CI: 0.9, 3.4) days, with comparable effect sizes across ethnicity (P for interaction=0.92). When considering offspring variants, similar results were observed for FADS3; each additional copy of the minor allele of the variant rs174450 was associated with a shorter gestation by 1.9 (0.7, 3.0) days. This association was driven by the Chinese and Indian subgroups, although the test for interaction with ethnicity was not significant (P for interaction=0.42). Offspring, but not maternal, FADS1 variant rs174546 (associated with higher LC-PUFA levels) was also associated with longer gestation duration [1.7 (0.3, 3.0) days per additional copy of the minor allele G], and this association was driven by the Chinese subgroup [2.4 (0.8, 4.1) days]. The other FADS1 variants and the FADS2 variants (maternal and fetal) were not associated with gestation duration.

Table 2.

Mean differences in gestation duration according to FADS gene variants in mother-offspring pairs from the GUSTO cohort.1

SNPs Minor allele All ethnic subgroups2 Chinese Malay Indian p-value for interaction3
Mothers n=502 n=284 n=150 n=68
    rs1745464 (FADS1) G 0.8 (-0.5, 2.1) 1.4 (-0.3, 3.1) -0.6 (-2.6, 1.5) 0.2 (-6.5, 6.8) 0.43
    rs2727270 (FADS2) A -0.5 (-1.8, 0.8) -1.3 (-3.0, 0.4) 1.1 (-0.8, 3.0) 0.1 (-8.8, 9.0) 0.24
    rs174593 (FADS2) G 0.3 (-1.3, 1.9) 0.5 (-1.8, 2.9) -0.8 (-2.8, 1.2) 4.3 (-1.7, 10.3) 0.23
    rs498793 (FADS2) A -0.1 (-1.8, 1.6) 1.4 (-1.6, 4.4) -0.5 (-2.6, 1.7) -2.0 (-6.0, 2.1) 0.33
    rs17156506 (FADS2) A 0.3 (-1.8, 2.4) -0.5 (-3.4, 2.5) 0.7 (-2.1, 3.4) 6.4 (-3.9, 16.6) 0.35
    rs174450 (FADS3) C -2.2 (-3.4, -0.9)*** -2.1 (-3.8, -0.4)* -2.1 (-4.0, -0.3)* -2.9 (-7.2, 1.4) 0.92
    rs1000778 (FADS3) A -1.6 (-2.8, -0.3)* -1.7 (-3.4, 0.1) -1.4 (-3.2, 0.4) -1.6 (-5.4, 2.3) 0.99
    rs174455 (FADS3) G -1.9 (-3.1, -0.8)** -1.9 (-3.6, -0.3)* -1.8 (-3.6, -0.1)* -2.4 (-6.3, 1.6) 0.97
Offspring n=601 n=337 n=174 n=90
    rs1745464 (FADS1) G 1.7 (0.3, 3.0)* 2.4 (0.8, 4.1)** 0.3 (-1.7, 2.4) -0.3 (-7.4, 6.8) 0.31
    rs2727270 (FADS2) A -1.0 (-2.3, 0.3) -2.0 (-3.7, -0.3)* 0.6 (-1.1, 2.2) -1.1 (-13.7, 11.6) 0.18
    rs174593 (FADS2) G -0.7 (-2.1, 0.8) -1.5 (-3.6, 0.6) -0.8 (-2.5, 0.9) 3.3 (-2.5, 9.2) 0.14
    rs498793 (FADS2) A 0.6 (-1.1, 2.2) 0.3 (-2.4, 2.9) 0.7 (-1.4, 2.9) 0.7 (-3.8, 5.2) 0.97
    rs17156506 (FADS2) A 0.8 (-1.3, 3.0) 0.3 (-2.5, 3.1) 0.8 (-1.9, 3.4) 6.3 (-5.3, 17.8) 0.43
    rs174450 (FADS3) C -1.9 (-3.0, -0.7)** -2.4 (-4.0, -0.9)** -0.7 (-2.4, 0.9) -2.2 (-6.1, 1.8) 0.42
    rs1000778 (FADS3) A -2.0 (-3.3, -0.8)** -2.4 (-4.1, -0.7)** -1.1 (-2.9, 0.6) -2.8 (-7.0, 1.5) 0.60
    rs174455 (FADS3) G -1.9 (-3.1, -0.8)** -2.8 (-4.3, -1.2)*** -0.5 (-2.2, 1.1) -2.1 (-6.1, 2.0) 0.25
Open in a new tab
1

Values are mean differences (95% CI) in gestation duration (in days) per copy of the minor allele among participants who experienced a spontaneous labor. Estimates were assessed by linear regression. p-values: *<0.05, **<0.01, ***<0.001.

2

Models in all ethnic subgroups were adjusted for ethnicity.

3

p-value for interaction of the SNP with ethnicity.

4

rs174546 is in negative linkage disequilibrium with the other SNPs.

In the cause-specific hazard model considering induced labor and pre-labor caesarean section as competing events of spontaneous labor, mothers homozygous for the minor allele C of rs174450 (decreasing LC-PUFA levels) had an increased cause-specific hazard ratio (csHR (95% CI): 1.51 (1.18, 1.95)) of delivery compared to mothers homozygous for the major allele A (Table 3). This was consistent across ethnic groups: 1.44 (1.03, 2.03), 1.46 (0.90, 2.39) and 1.48 (1.08, 2.02) in Chinese, Malay and Indian, respectively. When considering the offspring variants, comparable results were found with FADS3 variants in the overall population (1.51 (1.20, 1.89) in minor allele homozygotes for rs174450; Figure 2); this association was, however, driven by the Chinese subgroup [1.80 (1.31, 2.48)]. No associations were observed with FADS1 and FADS2 variants in non-ethnic-stratified analyses. The Fine–Gray subdistribution hazard models yielded comparable results (Supplemental Table 6), although with this model, the assumption of proportional hazards was mildly violated; in sensitivity analyses allowing for time-varying effects resulted in similar associations (not shown). Overall, there was no evidence of associations between FADS variants and the competing events, with the exception of induced labor which was associated with maternal FADS2 rs17156506 [csHR (95%): 2.90 (1.19, 7.08)] (Supplemental Table 7).

Table 3.

Cause-specific hazard ratios for spontaneous delivery according to FADS gene variants in mother-offspring pairs from the GUSTO cohort.1

All ethnic subgroups2 Chinese Malay Indian
SNPs Minor allele Mm mm Mm mm Mm mm Mm + mm3
Mothers n=897 n=516 n=231 n=150
    rs1745464 (FADS1) G 0.89 (0.73, 1.09) 0.82 (0.62, 1.08) 0.82 (0.63, 1.06) 0.70 (0.50, 0.99)* 1.08 (0.77, 1.51) 1.06 (0.55, 2.05) 1.06 (0.79, 1.43)
    rs2727270 (FADS2) A 1.02 (0.83, 1.26) 1.14 (0.88, 1.49) 1.09 (0.83, 1.43) 1.44 (1.04, 1.99)* 0.93 (0.64, 1.35) 0.78 (0.48, 1.26) 0.89 (0.60, 1.32)
    rs174593 (FADS2) G 1.16 (0.96, 1.41) 0.97 (0.61, 1.55) 0.99 (0.76, 1.30) 1.17 (0.55, 2.48) 1.36 (0.97, 1.91) 0.97 (0.54, 1.76) 1.18 (0.90, 1.55)
    rs498793 (FADS2) A 0.94 (0.76, 1.17) 1.48 (0.90, 2.45) 0.81 (0.58, 1.12) 1.85 (0.59, 5.79) 1.13 (0.79, 1.61) 3.42 (1.64, 7.11)** 0.87 (0.68, 1.11)
    rs17156506 (FADS2) A 1.17 (0.92, 1.50) 1.49 (0.66, 3.34) 1.31 (0.93, 1.84) 1.97 (0.73, 5.33) 1.04 (0.71, 1.52) 1.12 (0.28, 4.56) 1.04 (0.66, 1.64)
    rs174450 (FADS3) C 1.11 (0.90, 1.37) 1.51 (1.18, 1.95)** 1.06 (0.82, 1.38) 1.44 (1.03, 2.03)* 0.95 (0.58, 1.53) 1.46 (0.90, 2.39) 1.48 (1.08, 2.02)*
    rs1000778 (FADS3) A 1.04 (0.86, 1.25) 1.35 (1.03, 1.78)* 0.93 (0.73, 1.20) 1.46 (0.98, 2.18) 1.09 (0.77, 1.54) 1.25 (0.77, 2.04) 1.38 (1.05, 1.83)*
    rs174455 (FADS3) G 1.08 (0.88, 1.33) 1.42 (1.11, 1.82)** 1.04 (0.80, 1.35) 1.33 (0.95, 1.85) 0.97 (0.63, 1.49) 1.40 (0.89, 2.2) 1.46 (1.08, 1.99)*
Offspring n=1100 n=626 n=277 n=197
    rs1745464 (FADS1) G 0.94 (0.78, 1.14) 0.77 (0.59, 1.01) 0.96 (0.76, 1.22) 0.66 (0.47, 0.93)* 0.94 (0.69, 1.29) 1.04 (0.53, 2.06) 1.05 (0.80, 1.39)
    rs2727270 (FADS2) A 1.06 (0.87, 1.29) 1.10 (0.85, 1.42) 1.28 (1.00, 1.65) 1.20 (0.85, 1.68) 0.80 (0.56, 1.14) 0.95 (0.63, 1.42) 0.78 (0.50, 1.23)
    rs174593 (FADS2) G 1.07 (0.89, 1.28) 1.33 (0.91, 1.93) 1.13 (0.88, 1.44) 1.87 (0.92, 3.79) 0.93 (0.68, 1.29) 1.16 (0.73, 1.86) 1.08 (0.85, 1.37)
    rs498793 (FADS2) A 1.04 (0.85, 1.27) 1.07 (0.68, 1.69) 1.03 (0.77, 1.39) 1.61 (0.60, 4.34) 1.08 (0.77, 1.50) 1.26 (0.61, 2.58) 0.94 (0.76, 1.16)
    rs17156506 (FADS2) A 1.10 (0.88, 1.37) 2.41 (0.89, 6.48) 1.18 (0.89, 1.58) 3.89 (0.54, 27.77) 0.94 (0.64, 1.38) 1.85 (0.59, 5.81) 1.11 (0.73, 1.68)
    rs174450 (FADS3) C 1.22 (1.01, 1.48)* 1.51 (1.20, 1.89)*** 1.27 (0.99, 1.61) 1.80 (1.31, 2.48)*** 1.29 (0.86, 1.94) 1.45 (0.96, 2.18) 0.94 (0.74, 1.20)
    rs1000778 (FADS3) A 1.15 (0.97, 1.36) 1.28 (0.98, 1.68) 1.10 (0.88, 1.37) 1.64 (1.10, 2.44)* 1.20 (0.87, 1.66) 1.12 (0.70, 1.77) 1.13 (0.89, 1.43)
    rs174455 (FADS3) G 1.17 (0.97, 1.42) 1.40 (1.11, 1.77)** 1.20 (0.95, 1.52) 1.65 (1.19, 2.30)** 1.29 (0.87, 1.92) 1.39 (0.93, 2.09) 0.94 (0.74, 1.19)
Open in a new tab
1

Values are cause-specific hazard ratios (95% CI) for the incidence of birth by spontaneous labor for heterozygotes Mm and minor allele homozygotes mm (as compared to major allele homozygotes MM). Estimates were assessed by cause-specific proportional hazards model. p-values: *<0.05, **<0.01, ***<0.001.

2

Models in all ethnic subgroups were adjusted for ethnicity.

3

In Indians, homozygotes for the minor allele were too few and were therefore combined with the heterozygote group.

4

rs174546 is in negative linkage disequilibrium with the other SNPs.

Figure 2.

Figure 2

Open in a new tab

Cumulative incidence of the event of delivery after spontaneous labor according to offspring FADS3 rs174450 variant (major and minor alleles are A and C, respectively). Cumulative incidence function estimates were derived from cause-specific proportional hazards model adjusted for ethnicity (n=1100).

Finally, FADS1 and FADS3 genes variants were associated with crude birth length and weight among participants who experienced spontaneous labor. However, this association was almost entirely removed after accounting for gestation duration (Table 4).

Table 4.

Mean differences in birth length and weight according to FADS gene variants in mother-offspring pairs from the GUSTO cohort.1

Birth length Birth weight
SNPs Minor allele Unadjusted2, cm Gestational age-specific z-score Unadjusted2, g Gestational age-specific z-score
Mothers n=502 n=502
    rs1745463 (FADS1) G 0.33 (0.03, 0.62)* 0.10 (-0.04, 0.23) 47 (-10, 104) 0.06 (-0.08, 0.19)
    rs2727270 (FADS2) A -0.34 (-0.63, -0.05)* -0.11 (-0.25, 0.02) -42 (-98, 14) -0.06 (-0.19, 0.07)
    rs174593 (FADS2) G 0.02 (-0.34, 0.37) 0.01 (-0.15, 0.17) -3 (-71, 66) -0.02 (-0.18, 0.14)
    rs498793 (FADS2) A 0.24 (-0.14, 0.62) 0.15 (-0.02, 0.33) 73 (0, 147) 0.24 (0.07, 0.41)**
    rs17156506 (FADS2) A -0.52 (-0.99, -0.05)* -0.30 (-0.52, -0.08)** -58 (-149, 33) -0.16 (-0.38, 0.05)
    rs174450 (FADS3) C -0.35 (-0.63, -0.07)* 0.00 (-0.13, 0.13) -66 (-120, -12)* 0.00 (-0.13, 0.12)
    rs1000778 (FADS3) A -0.17 (-0.45, 0.11) 0.05 (-0.08, 0.18) -45 (-99, 9) -0.01 (-0.13, 0.12)
    rs174455 (FADS3) G -0.37 (-0.63, -0.10)** -0.03 (-0.15, 0.10) -75 (-127, -23)** -0.05 (-0.17, 0.07)
Offspring n=601 n=601
    rs1745463 (FADS1) G 0.52 (0.22, 0.81)*** 0.15 (0.02, 0.28)* 66 (9, 122)* 0.05 (-0.08, 0.18)
    rs2727270 (FADS2) A -0.28 (-0.57, 0.00) -0.05 (-0.17, 0.08) -32 (-86, 22) 0.02 (-0.10, 0.15)
    rs174593 (FADS2) G -0.31 (-0.62, 0.01) -0.16 (-0.30, -0.02)* -34 (-94, 27) -0.09 (-0.23, 0.05)
    rs498793 (FADS2) A 0.13 (-0.23, 0.49) 0.03 (-0.13, 0.19) 63 (-5, 131) 0.14 (-0.02, 0.29)
    rs17156506 (FADS2) A -0.34 (-0.80, 0.12) -0.20 (-0.41, 0.00) -33 (-121, 55) -0.09 (-0.30, 0.11)
    rs174450 (FADS3) C -0.48 (-0.73, -0.23)*** -0.13 (-0.25, -0.02)* -63 (-111, -15)* -0.05 (-0.16, 0.06)
    rs1000778 (FADS3) A -0.33 (-0.60, -0.06)* -0.03 (-0.15, 0.09) -57 (-109, -5)* -0.04 (-0.16, 0.08)
    rs174455 (FADS3) G -0.43 (-0.68, -0.17)** -0.10 (-0.21, 0.02) -77 (-125, -28)** -0.09 (-0.20, 0.02)
Open in a new tab
1

Values are mean differences (95% CI) in birth length and weight per copy of the minor allele among participants who experienced a spontaneous labor. Estimates were assessed by linear regression. p-values: *<0.05, **<0.01, ***<0.001.

2

Models were adjusted for ethnicity.

3

rs174546 is in negative linkage disequilibrium with the other SNPs.

Discussion

In our tri-ethnic mother-offspring cohort, maternal FADS variants were associated with maternal ω6 but not ω3 PUFA levels. Among the mothers who experienced spontaneous labor, FADS3 variants rs174450, rs1000778 and rs174455 were associated with gestation duration, with similar magnitudes across ethnicities. Among offspring born after spontaneous labor, FADS1 variant rs174546 and all FADS3 variants were also associated with gestation duration, but less consistently so across ethnicities. In survival models considering induced labor and pre-labor caesarean section as competing events, FADS3 variants were similarly related to time to delivery after spontaneous labor. Finally, maternal and offspring FADS1 variant rs174546 and FADS3 variants were associated with length and weight at birth, but these associations were almost entirely removed after accounting for gestation duration.

Our study includes several novel features. First, ours is the first MR study of associations between genetic variants involved in PUFA metabolism and gestation duration and birth size. In offspring, minor allele of FADS1 variant rs174546, which is in high linkage disequilibrium with eight other variants located on FADS1 and FADS2 (Supplemental Table 1), was associated with longer gestation duration. This finding supports our initial hypothesis and adds new evidence to published meta-analyses of RCTs and observational studies reporting that exposure to LC-PUFAs during pregnancy is associated with duration of gestation (17–19). Our findings also suggest that most of the effect observed on birth size is mediated by gestation duration. However, our study falls short in demonstrating a causal role of DHA on gestation duration because FADS1 variants were not associated with plasma DHA levels in our cohort, which is in line with recently published genome-wide association studies in Chinese populations (29, 30). A strength of the MR design is to limit potential residual confounding that could arise from the different genetic background and dietary pattern between the three ethnic groups. A second strength of MR is to avoid potential reverse causality issues, although we cannot exclude the possibility that FADS variants lead to differences in growth very early in gestation and thus to a differential estimate of gestational age on the dating scan (35).

A second novel aspect of our study is the consistent association of FADS3 variants with pregnancy outcomes. The function of FADS3 remains largely unknown but is presumed to be involved in PUFA metabolism (11, 12). It is expressed in many brain areas in baboons and may be expressed in the human placenta (36, 37). It is well known that LC-PUFAs are required by the developing brain. During Homo sapiens’s speciation process, an evolutionary trade-off may have occurred to balance the constraints imposed by bipedalism (shorter maternal pelvic region) and encephalization (larger newborn head), thereby increasing maternal mortality. One of these trade-offs may have been shortening gestation and hence reducing the physiological maturity of the offspring at birth compared to other primates, thus leading to increased infant mortality (38). In this context, it is probable that the FADS variants maximizing survival of both the mother and the infant have been positively selected throughout human evolution. The fact that human breast milk contains higher levels of the ω3 LC-PUFA DHA than that of other mammals, including the other primates, corroborates such an evolutionary hypothesis (39). Other population genetics studies have shown the FADS gene cluster has been under strong selection pressure and may even have played a key role in evolutionary adaptation of modern humans (40–44). We speculate that the FADS genes cluster, including FADS3, could play enhanced roles during pregnancy, either via placental mechanisms or direct effects on the developing fetus. Mechanistic studies on placental and fetal tissues are required to further explore these speculations.

Our results related to FADS variants add new insights into how LC-PUFAs may affect gestation duration. Previous studies have found that ω3 LC-PUFA supplementation prolongs gestation duration. Our findings suggest an additional role of ω6 LC-PUFAs. AA is the precursor of prostaglandins, especially prostaglandin E2, which is involved in cervical ripening and labor initiation and is often administered vaginally to induce labor (45). Animal models report that supplementation with EPA- and DHA-rich food reduces AA metabolite synthesis and might thereby delay labor onset (46). A balance of ω3 and ω6 LC-PUFAs may influence the pathways involved in optimal gestation duration. In our study, LC-PUFA status was measured at mid-pregnancy, which may be a poor proxy for LC-PUFA status at delivery. ω3 LC-PUFA status is also more influenced by dietary intake than by FADS variants, thereby weakening those variants as causal instruments in the MR design, particularly in Chinese populations as compared to European populations (29). Using the formula from Brion et al. (47), we calculated that >300,000 participants would been necessary to detect an effect of DHA on gestation duration using rs174546 instrument in our study population. Future longitudinal measurements of AA metabolites in maternal blood, placenta and cord blood should provide a better understanding of the mechanisms differentially regulated by ω3 and ω6 LC-PUFAs. An additional area of investigation will be the competition between the mother and the fetus for sharing limited resources of LC-PUFA. This will help address the issue of supply and demand and its impact on gestation duration and fetal growth.

The main limitation of our study is that FADS variants were, in our setting, poor causal instruments of ω3 LC-PUFA status. Additional limitations must be noted. First, MR assumption that there is no horizontal pleiotropy cannot be assessed (48). FADS variants might have effects on gestation duration through pathways other than LC-PUFA status. The known determinants of gestation duration are, however, unlikely to be caused by variation in FADS genes (49). Second, our sample size was small compared to current standards in genetic epidemiology, and we must therefore lack statistical power, especially when stratifying by ethnic subgroup. Nonetheless, we found no interactions with ethnicity in the full sample, and most observed associations were consistent across ethnic subgroups, thus supporting an ethnic-adjusted pooled approach. Another limitation is that the assumption of proportional hazards was mildly violated in the Fine–Gray subdistribution hazard models. However, results from Fine–Gray analyses, even when misspecified, still provide a useful summary measure approximating the true effect size (50). Furthermore, the results with and without time-varying effects were consistent with those obtained by linear regression restricted to women who experienced spontaneous labor.

Implications of our study are both etiological and clinical. Our findings add further evidence to pre-existing demonstrations of a causal effect of LC-PUFAs on gestation duration. This knowledge may also help improve clinical practices to reduce the risks related to short gestation. Minor allele carriers of FADS3 variants appear to have an earlier delivery, with an average effect size of 4 (2 to 6) days between major allele homozygotes and minor allele homozygotes. Prolonging gestation to this extent can be of clinical importance for fetal development, and hence for postnatal care and health care costs.

In conclusion, using an MR approach for FADS cluster variants, we add new evidence that maternal LC-PUFA status during pregnancy affects gestation duration. In agreement with other studies, we also found that the observed larger size at birth is mainly attributable to longer gestation. We found consistent associations of FADS3 variants with gestation duration, which are novel and suggest a role for the FADS3 gene in placental function and/or fetal development. Our study calls for replication of MR design in larger independent epidemiological studies. Further mechanistic work is needed to understand the underlying biological mechanisms.

Supplementary Material

Supplemental Material

Acknowledgement

The GUSTO study group includes Allan Sheppard, Amutha Chinnadurai, Anne Eng Neo Goh, Anne Rifkin-Graboi, Anqi Qiu, Arijit Biswas, Bee Wah Lee, Birit F.P. Broekman, Boon Long Quah, Borys Shuter, Chai Kiat Chng, Cheryl Ngo, Choon Looi Bong, Christiani Jeyakumar Henry, Cornelia Yin Ing Chee, Yam Thiam Daniel Goh, Doris Fok, Fabian Yap, George Seow Heong Yeo, Helen Chen, Hugo P S van Bever, Iliana Magiati, Inez Bik Yun Wong, Ivy Yee-Man Lau, Jeevesh Kapur, Jenny L. Richmond, Jerry Kok Yen Chan, Joanna D. Holbrook, Joshua J. Gooley, Keith M. Godfrey, Kenneth Kwek, Kok Hian Tan, Krishnamoorthy Niduvaje, Leher Singh, Lin Lin Su, Lourdes Mary Daniel, Lynette P Shek, Marielle V. Fortier, Mark Hanson, Mary Foong-Fong Chong, Mary Rauff, Mei Chien Chua, Michael Meaney, Mya Thway Tint, Neerja Karnani, Ngee Lek, Oon Hoe Teoh, P. C. Wong, Peter D. Gluckman, Pratibha Agarwal, Rob M. van Dam, Salome A. Rebello, Seang-Mei Saw, Shang Chee Chong, Shirong Cai, Shu-E Soh, Sok Bee Lim, Chin-Ying Stephen Hsu, Victor Samuel Rajadurai, Walter Stunkel, Wee Meng Han, Wei Wei Pang, Yap-Seng Chong, Yin Bun Cheung, Yiong Huak Chan and Yung Seng Lee.

Sources of Support: This work was supported by the Singapore National Research Foundation under its Translational and Clinical Research (TCR) Flagship Programme and administered by the Singapore Ministry of Health’s National Medical Research Council (NMRC) [NMRC/TCR/004-NUS/2008, NMRC/TCR/012-NUHS/2014]; Singapore Institute for Clinical Sciences; Agency for Science Technology and Research (A*STAR) and Nestec. The study sponsors were not involved in study design, data collection, analysis and interpretation of data, manuscript writing, and the decision to submit the article for publication. PCC and KMG are supported by the National Institute for Health Research through the NIHR Southampton Biomedical Research Centre and KMG is supported by the European Union's Seventh Framework Programme (FP7/2007-2013), project EarlyNutrition [grant agreement number 289346].

Abbreviations

GUSTO

growing up in Singapore towards healthy outcomes

ALA

alpha-linolenic acid

EPA

eicosapentaenoic acid

DHA

docosahexaenoic acid

LA

linoleic acid

AA

arachidonic acid

PUFA

polyunsaturated fatty acid

LC-PUFA

long-chain polyunsaturated fatty acid

FADS1

fatty acid desaturase 1

FADS2

fatty acid desaturase 2

FADS3

fatty acid desaturase 3

SNP

single nucleotide polymorphism

MR

Mendelian randomization.

Footnotes

Clinical Trial Registration Number: NCT01174875 on clinicaltrials.gov.

Conflict of Interest Statement: LS, YSC, PCC and KMG have received reimbursement for speaking at conferences sponsored by companies selling nutritional products. They are part of an academic consortium that has received research funding from Abbott Nutrition, Nestle, and Danone. The other authors have no potential conflicts of interest to disclose.

Authors’ Contributions:

JYB, FY, KHT, LS, YSC, PDG, KMG, MFFC, MSK, NK and YSL designed research; HP, IMA, SES and PCC conducted research; JYB, HP, IMA, SES and MFFC provided essential materials; JYB and MMB performed statistical analysis, JYB, MMB and MSK wrote paper; JYB had primary responsibility for final content.

References

  • 1.Koletzko B, Lien E, Agostoni C, Bohles H, Campoy C, Cetin I, Decsi T, Dudenhausen JW, Dupont C, Forsyth S, et al. The roles of long-chain polyunsaturated fatty acids in pregnancy, lactation and infancy: review of current knowledge and consensus recommendations. J Perinat Med. 2008;36(1):5–14. doi: 10.1515/JPM.2008.001. [DOI] [PubMed] [Google Scholar]
  • 2.Clandinin MT, Chappell JE, Leong S, Heim T, Swyer PR, Chance GW. Intrauterine fatty acid accretion rates in human brain: implications for fatty acid requirements. Early Hum Dev. 1980;4(2):121–9. doi: 10.1016/0378-3782(80)90015-8. [DOI] [PubMed] [Google Scholar]
  • 3.Brenna JT, Carlson SE. Docosahexaenoic acid and human brain development: evidence that a dietary supply is needed for optimal development. Journal of human evolution. 2014;77:99–106. doi: 10.1016/j.jhevol.2014.02.017. [DOI] [PubMed] [Google Scholar]
  • 4.Gil-Sanchez A, Koletzko B, Larque E. Current understanding of placental fatty acid transport. Curr Opin Clin Nutr Metab Care. 2012;15(3):265–72. doi: 10.1097/MCO.0b013e3283523b6e. [DOI] [PubMed] [Google Scholar]
  • 5.Meyer BJ, Mann NJ, Lewis JL, Milligan GC, Sinclair AJ, Howe PR. Dietary intakes and food sources of omega-6 and omega-3 polyunsaturated fatty acids. Lipids. 2003;38(4):391–8. doi: 10.1007/s11745-003-1074-0. [DOI] [PubMed] [Google Scholar]
  • 6.Plourde M, Cunnane SC. Extremely limited synthesis of long chain polyunsaturates in adults: implications for their dietary essentiality and use as supplements. Appl Physiol Nutr Metab. 2007;32(4):619–34. doi: 10.1139/H07-034. [DOI] [PubMed] [Google Scholar]
  • 7.Burdge GC, Wootton SA. Conversion of alpha-linolenic acid to eicosapentaenoic, docosapentaenoic and docosahexaenoic acids in young women. Br J Nutr. 2002;88(4):411–20. doi: 10.1079/BJN2002689. [DOI] [PubMed] [Google Scholar]
  • 8.Childs CE, Romeu-Nadal M, Burdge GC, Calder PC. Gender differences in the n-3 fatty acid content of tissues. Proc Nutr Soc. 2008;67(1):19–27. doi: 10.1017/S0029665108005983. [DOI] [PubMed] [Google Scholar]
  • 9.Walker CG, Browning LM, Mander AP, Madden J, West AL, Calder PC, Jebb SA. Age and sex differences in the incorporation of EPA and DHA into plasma fractions, cells and adipose tissue in humans. Br J Nutr. 2014;111(4):679–89. doi: 10.1017/S0007114513002985. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Schaeffer L, Gohlke H, Muller M, Heid IM, Palmer LJ, Kompauer I, Demmelmair H, Illig T, Koletzko B, Heinrich J. Common genetic variants of the FADS1 FADS2 gene cluster and their reconstructed haplotypes are associated with the fatty acid composition in phospholipids. Hum Mol Genet. 2006;15(11):1745–56. doi: 10.1093/hmg/ddl117. [DOI] [PubMed] [Google Scholar]
  • 11.Tosi F, Sartori F, Guarini P, Olivieri O, Martinelli N. Delta-5 and delta-6 desaturases: crucial enzymes in polyunsaturated fatty acid-related pathways with pleiotropic influences in health and disease. Adv Exp Med Biol. 2014;824:61–81. doi: 10.1007/978-3-319-07320-0_7. [DOI] [PubMed] [Google Scholar]
  • 12.Blanchard H, Legrand P, Pedrono F. Fatty Acid Desaturase 3 (Fads3) is a singular member of the Fads cluster. Biochimie. 2011;93(1):87–90. doi: 10.1016/j.biochi.2010.03.002. [DOI] [PubMed] [Google Scholar]
  • 13.Fahmida U, Htet MK, Adhiyanto C, Kolopaking R, Yudisti MA, Maududi A, Suryandari DA, Dillon D, Afman L, Muller M. Genetic variants of FADS gene cluster, plasma LC-PUFA levels and the association with cognitive function of under-two-year-old Sasaknese Indonesian children. Asia Pac J Clin Nutr. 2015;24(2):323–8. doi: 10.6133/apjcn.2015.24.2.17. [DOI] [PubMed] [Google Scholar]
  • 14.Koletzko B, Lattka E, Zeilinger S, Illig T, Steer C. Genetic variants of the fatty acid desaturase gene cluster predict amounts of red blood cell docosahexaenoic and other polyunsaturated fatty acids in pregnant women: findings from the Avon Longitudinal Study of Parents and Children. Am J Clin Nutr. 2011;93(1):211–9. doi: 10.3945/ajcn.110.006189. [DOI] [PubMed] [Google Scholar]
  • 15.Lattka E, Rzehak P, Szabo E, Jakobik V, Weck M, Weyermann M, Grallert H, Rothenbacher D, Heinrich J, Brenner H, et al. Genetic variants in the FADS gene cluster are associated with arachidonic acid concentrations of human breast milk at 1.5 and 6 mo postpartum and influence the course of milk dodecanoic, tetracosenoic, and trans-9-octadecenoic acid concentrations over the duration of lactation. Am J Clin Nutr. 2011;93(2):382–91. doi: 10.3945/ajcn.110.004515. [DOI] [PubMed] [Google Scholar]
  • 16.Morales E, Bustamante M, Gonzalez JR, Guxens M, Torrent M, Mendez M, Garcia-Esteban R, Julvez J, Forns J, Vrijheid M, et al. Genetic variants of the FADS gene cluster and ELOVL gene family, colostrums LC-PUFA levels, breastfeeding, and child cognition. PLoS One. 2011;6(2):e17181. doi: 10.1371/journal.pone.0017181. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Szajewska H, Horvath A, Koletzko B. Effect of n-3 long-chain polyunsaturated fatty acid supplementation of women with low-risk pregnancies on pregnancy outcomes and growth measures at birth: a meta-analysis of randomized controlled trials. Am J Clin Nutr. 2006;83(6):1337–44. doi: 10.1093/ajcn/83.6.1337. [DOI] [PubMed] [Google Scholar]
  • 18.Makrides M, Duley L, Olsen SF. Marine oil, and other prostaglandin precursor, supplementation for pregnancy uncomplicated by pre-eclampsia or intrauterine growth restriction. Cochrane Database Syst Rev. 2006(3):CD003402. doi: 10.1002/14651858.CD003402.pub2. [DOI] [PubMed] [Google Scholar]
  • 19.Horvath A, Koletzko B, Szajewska H. Effect of supplementation of women in high-risk pregnancies with long-chain polyunsaturated fatty acids on pregnancy outcomes and growth measures at birth: a meta-analysis of randomized controlled trials. Br J Nutr. 2007;98(2):253–9. doi: 10.1017/S0007114507709078. [DOI] [PubMed] [Google Scholar]
  • 20.Olsen SF. Low consumption of seafood in early pregnancy as a risk factor for preterm delivery: prospective cohort study. BMJ. 2002;324(7335):447–50. doi: 10.1136/bmj.324.7335.447. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Oken E, Kleinman KP, Olsen SF, Rich-Edwards JW, Gillman MW. Associations of seafood and elongated n-3 fatty acid intake with fetal growth and length of gestation: results from a US pregnancy cohort. Am J Epidemiol. 2004;160(8):774–83. doi: 10.1093/aje/kwh282. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Leventakou V, Roumeliotaki T, Martinez D, Barros H, Brantsaeter AL, Casas M, Charles MA, Cordier S, Eggesbo M, van Eijsden M, et al. Fish intake during pregnancy, fetal growth, and gestational length in 19 European birth cohort studies. Am J Clin Nutr. 2014;99(3):506–16. doi: 10.3945/ajcn.113.067421. [DOI] [PubMed] [Google Scholar]
  • 23.Mansournia MA, Higgins JP, Sterne JA, Hernan MA. Biases in Randomized Trials: A Conversation Between Trialists and Epidemiologists. Epidemiology. 2017;28(1):54–9. doi: 10.1097/EDE.0000000000000564. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Hernán MA, Robins JM. Causal inference. Boca Raton: Chapman & Hall/CRC; 2018. [Google Scholar]
  • 25.Krieger N, Davey Smith G. The tale wagged by the DAG: broadening the scope of causal inference and explanation for epidemiology. Int J Epidemiol. 2016 doi: 10.1093/ije/dyw114. [DOI] [PubMed] [Google Scholar]
  • 26.Davey Smith G, Ebrahim S. ‘Mendelian randomization’: can genetic epidemiology contribute to understanding environmental determinants of disease?*. Int J Epidemiol. 2003;32(1):1–22. doi: 10.1093/ije/dyg070. [DOI] [PubMed] [Google Scholar]
  • 27.Soh SE, Tint MT, Gluckman PD, Godfrey KM, Rifkin-Graboi A, Chan YH, Stunkel W, Holbrook JD, Kwek K, Chong YS, et al. Cohort profile: Growing Up in Singapore Towards healthy Outcomes (GUSTO) birth cohort study. Int J Epidemiol. 2014;43(5):1401–9. doi: 10.1093/ije/dyt125. [DOI] [PubMed] [Google Scholar]
  • 28.Chong MF, Ong YL, Calder PC, Colega M, Wong JX, Tan CS, Lim AL, Fisk HL, Cai S, Pang WW, et al. Long-chain polyunsaturated fatty acid status during pregnancy and maternal mental health in pregnancy and the postpartum period: results from the GUSTO study. J Clin Psychiatry. 2015;76(7):e848–56. doi: 10.4088/JCP.14m09191. [DOI] [PubMed] [Google Scholar]
  • 29.Hu Y, Li H, Lu L, Manichaikul A, Zhu J, Chen YD, Sun L, Liang S, Siscovick DS, Steffen LM, et al. Genome-wide meta-analyses identify novel loci associated with n-3 and n-6 polyunsaturated fatty acid levels in Chinese and European-ancestry populations. Hum Mol Genet. 2016;25(6):1215–24. doi: 10.1093/hmg/ddw002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Dorajoo R, Sun Y, Han Y, Ke T, Burger A, Chang X, Low HQ, Guan W, Lemaitre RN, Khor CC, et al. A genome-wide association study of n-3 and n-6 plasma fatty acids in a Singaporean Chinese population. Genes Nutr. 2015;10(6):53. doi: 10.1007/s12263-015-0502-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Aris IM, Gandhi M, Cheung YB, Soh SE, Tint MT, Gluckman PD, Lee YS, Yap FK, Chong YS. A new population-based reference for gestational age-specific size-at-birth of Singapore infants. Ann Acad Med Singapore. 2014;43(9):439–47. [PubMed] [Google Scholar]
  • 32.Lau B, Cole SR, Gange SJ. Competing risk regression models for epidemiologic data. Am J Epidemiol. 2009;170(2):244–56. doi: 10.1093/aje/kwp107. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Latouche A, Allignol A, Beyersmann J, Labopin M, Fine JP. A competing risks analysis should report results on all cause-specific hazards and cumulative incidence functions. J Clin Epidemiol. 2013;66(6):648–53. doi: 10.1016/j.jclinepi.2012.09.017. [DOI] [PubMed] [Google Scholar]
  • 34.Teh AL, Pan H, Chen L, Ong ML, Dogra S, Wong J, MacIsaac JL, Mah SM, McEwen LM, Saw SM, et al. The effect of genotype and in utero environment on interindividual variation in neonate DNA methylomes. Genome Res. 2014;24(7):1064–74. doi: 10.1101/gr.171439.113. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Slama R, Khoshnood B, Kaminski M. How to control for gestational age in studies involving environmental effects on fetal growth. Environ Health Perspect. 2008;116(7):A284. doi: 10.1289/ehp.11105. author reply A-A5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Fagerberg L, Hallstrom BM, Oksvold P, Kampf C, Djureinovic D, Odeberg J, Habuka M, Tahmasebpoor S, Danielsson A, Edlund K, et al. Analysis of the human tissue-specific expression by genome-wide integration of transcriptomics and antibody-based proteomics. Mol Cell Proteomics. 2014;13(2):397–406. doi: 10.1074/mcp.M113.035600. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Mure LS, Le HD, Benegiamo G, Chang MW, Rios L, Jillani N, Ngotho M, Kariuki T, Dkhissi-Benyahya O, Cooper HM, et al. Diurnal transcriptome atlas of a primate across major neural and peripheral tissues. Science. 2018:eaao0318. doi: 10.1126/science.aao0318. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Diamond JM. The Third Chimpanzee: The Evolution and Future of the Human Animal. Harper Perennial; New York City, NYC, USA: 2006. [Google Scholar]
  • 39.Milligan LA, Rapoport SI, Cranfield MR, Dittus W, Glander KE, Oftedal OT, Power ML, Whittier CA, Bazinet RP. Fatty acid composition of wild anthropoid primate milks. Comp Biochem Physiol B Biochem Mol Biol. 2008;149(1):74–82. doi: 10.1016/j.cbpb.2007.08.006. [DOI] [PubMed] [Google Scholar]
  • 40.Mathias RA, Fu W, Akey JM, Ainsworth HC, Torgerson DG, Ruczinski I, Sergeant S, Barnes KC, Chilton FH. Adaptive evolution of the FADS gene cluster within Africa. PLoS One. 2012;7(9):e44926. doi: 10.1371/journal.pone.0044926. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Genomes Project C. Auton A, Brooks LD, Durbin RM, Garrison EP, Kang HM, Korbel JO, Marchini JL, McCarthy S, McVean GA, et al. A global reference for human genetic variation. Nature. 2015;526(7571):68–74. doi: 10.1038/nature15393. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 42.Fumagalli M, Moltke I, Grarup N, Racimo F, Bjerregaard P, Jorgensen ME, Korneliussen TS, Gerbault P, Skotte L, Linneberg A, et al. Greenlandic Inuit show genetic signatures of diet and climate adaptation. Science. 2015;349(6254):1343–7. doi: 10.1126/science.aab2319. [DOI] [PubMed] [Google Scholar]
  • 43.Kothapalli KS, Ye K, Gadgil MS, Carlson SE, O'Brien KO, Zhang JY, Park HG, Ojukwu K, Zou J, Hyon SS, et al. Positive Selection on a Regulatory Insertion-Deletion Polymorphism in FADS2 Influences Apparent Endogenous Synthesis of Arachidonic Acid. Mol Biol Evol. 2016 doi: 10.1093/molbev/msw049. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Amorim CG, Nunes K, Meyer D, Comas D, Bortolini MC, Salzano FM, Hunemeier T. Genetic signature of natural selection in first Americans. Proc Natl Acad Sci U S A. 2017;114(9):2195–9. doi: 10.1073/pnas.1620541114. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Thomas J, Fairclough A, Kavanagh J, Kelly AJ. Vaginal prostaglandin (PGE2 and PGF2a) for induction of labour at term. Cochrane Database Syst Rev. 2014;(6):CD003101. doi: 10.1002/14651858.CD003101.pub3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.LeBlanc CJ, Horohov DW, Bauer JE, Hosgood G, Mauldin GE. Effects of dietary supplementation with fish oil on in vivo production of inflammatory mediators in clinically normal dogs. Am J Vet Res. 2008;69(4):486–93. doi: 10.2460/ajvr.69.4.486. [DOI] [PubMed] [Google Scholar]
  • 47.Brion MJ, Shakhbazov K, Visscher PM. Calculating statistical power in Mendelian randomization studies. Int J Epidemiol. 2013;42(5):1497–501. doi: 10.1093/ije/dyt179. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Davey Smith G, Hemani G. Mendelian randomization: genetic anchors for causal inference in epidemiological studies. Hum Mol Genet. 2014;23(R1):R89–98. doi: 10.1093/hmg/ddu328. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Kramer MS. Intrauterine growth and gestational duration determinants. Pediatrics. 1987;80(4):502–11. [PubMed] [Google Scholar]
  • 50.Grambauer N, Schumacher M, Beyersmann J. Proportional subdistribution hazards modeling offers a summary analysis, even if misspecified. Stat Med. 2010;29(7–8):875–84. doi: 10.1002/sim.3786. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplemental Material
NIHMS78520-supplement-Supplemental_Material.docx (468.9KB, docx)

RESOURCES