Relationship Between Serum Polyunsaturated Fatty Acids and Pregnancy in Women Undergoing In Vitro Fertilization

Abstract

Context:

Polyunsaturated fatty acids (PUFAs) and their metabolism may be important in normal reproductive function and fertility. Associations between physiologic PUFAs and pregnancy have not been established in women.

Objective:

The purpose of this study was to investigate associations between serum levels of PUFAs and embryo implantation in women undergoing in vitro fertilization (IVF).

Design:

This was a prospective cohort study conducted between 2010 and 2012.

Setting:

The study was conducted at the Washington University Reproductive Medicine Center.

Patients:

Participants were 200 women undergoing IVF and participating in an ongoing specimen tissue bank.

Intervention:

Fasting serum PUFAs were measured with liquid chromatography-mass spectroscopy. PUFAs measured included linoleic acid (LA), α-linolenic acid (ALA), eicosapentaenoic acid, arachidonic acid, and docosahexaenoic acid.

Main Outcome Measures:

Relationships between serum levels of measured PUFAs and embryo implantation in women undergoing IVF were analyzed.

Results:

In unadjusted analyses, none of the PUFAs alone were associated with a chance of pregnancy; however, women with increased LA:ALA ratios had a higher chance of pregnancy compared with women with lower LA:ALA ratios (relative risk, 1.52; 95% confidence interval, 1.09–2.13). This relationship held after multivariable logistic regression adjusting for age, antral follicle count, body mass index, history of previous pregnancy, and history of endometriosis (odds ratio, 2.7; 95% confidence interval, 1.3–5.7). Embryo implantation rates were also weakly associated with LA:ALA ratios (r = 0.21, P = .003).

Conclusions:

Our work shows that increased ω-6 to ω-3 PUFA ratios in women undergoing IVF are associated with increased implantation and pregnancy rates. Prospective trials are needed to determine whether manipulation of PUFA ratios through diet or pharmacologic intervention may benefit women planning to conceive.

In vitro studies and animal models suggest that free fatty acids (FFAs) play an important role in embryo implantation (1–3); however, few studies have examined the relationship between pregnancy and circulating FFA concentrations. Polyunsaturated fatty acids (PUFAs) are FFAs containing 2 or more double bonds and are classified into ω-3 (n−3), ω-6 (n−6), or ω-9 (n−9). n−3 and n−6 PUFAs are derived from the diet, and dietary intake influences their availability (4). Linoleic acid (LA), an n−6 PUFA, and α-linolenic acid (ALA), an n−3 PUFA, are the primary PUFAs in Western diets (5). Upon consumption, n−6 and n−3 PUFAs are metabolized by the same desaturase and elongase enzymes. An excess of one can influence the availability of the other; thus, n−6 to n−3 PUFA ratios may also be important.

Previously, we conducted a prospective study analyzing serum FFAs and pregnancy in women undergoing in vitro fertilization (IVF) (6). We found that ALA levels were negatively associated with pregnancy. Our current study further investigates the relationships between serum PUFA levels and pregnancy in women undergoing IVF.

Materials and Methods

Study participants and samples

Washington University's Institutional Review Board approved all protocols. Women were enrolled in the Women's and Infant's Health Specimen Consortium (WIHSC), a tissue bank at Washington University. All women underwent controlled ovarian hyperstimulation as described previously (7). In brief, women underwent a long GnRH agonist protocol. Gonadotropin dosing was based primarily on age and adjusted by response. Patients received 10 000 IU of urinary human chorionic gonadotropin when ≥2 follicles of ≥18 mm were noted. Oocyte retrieval took place 36 hours later. Fasting venous blood was collected the morning of oocyte retrieval. Isolated serum samples were stored at −80°C for batch analysis. Women who underwent embryo transfer and consented to studies through the WIHSC were included. Women using testicular biopsy sperm were excluded. Clinical data included body mass index (BMI) (in kilograms per meter squared), infertility diagnosis, parity, age, and antral follicle count. IVF laboratory data and outcomes included peak estradiol level, day of embryo transfer, number of embryos transferred, and clinical pregnancy and live birth information. Clinical pregnancy was our primary outcome, defined as a gestational sac on ultrasound.

Serum FFA analysis

Serum FFAs were analyzed by liquid chromatography-mass spectroscopy as described previously (8). In brief, serum FFAs were extracted by a modified Bligh-Dyer protocol with d4-palmitic acid, an internal standard. Extracted FFAs were further derivatized by dimethylaminopropylamine into amides to maximize sensitivity. Liquid chromatography was performed with a Shimadzu 10A high-performance liquid chromatography system and a Phenomenex Luna C18 column. Mass spectroscopy was performed with a TSQ Quantum Ultra triple quadrupole mass spectrometer operated in selected reaction monitoring mode under positive electrospray ionization. The identity of FFAs was determined by comparing retention time to corresponding commercial standards. Data processing was conducted with Xcalibur 2.0.7 software (Thermo Fisher Scientific, Waltham, Massachusetts). FFA concentrations were calculated as the concentration of d4-palmitic acid multiplied by the peak area ratio of the analyte to the internal standard. The coefficient of variation (CV) for each FFA was calculated to evaluate the precision of measurement. The intra-assay CV was <9%, and the interassay CV was <6% for each FFA measured.

Statistical analyses

Sample size was based on our previous work showing a negative association between serum ALA levels and pregnancy in women younger than 38 years undergoing their first cycle of IVF (6). The current study was initiated to further explore associations between essential FFAs, the products of their metabolism, and IVF outcomes in a more general population of women. We assumed an 82% pregnancy rate in women with the lowest ALA levels and a 48% pregnancy rate in women with the highest ALA levels, .05 type I error, .1 type II error, and a 1:1 case-control ratio and determined that 180 patients were necessary to detect a difference in chance of pregnancy based on specific serum FFA levels. A total of 200 women in the WIHSC database met the study criteria.

Serum FFAs were analyzed and matched with clinical data. Women were divided into 2 groups, those who conceived and those who did not. Appropriate parametric and nonparametric statistics were used to determine associations between baseline patient characteristics and pregnancy. Standard regression was used to determine associations between specific serum FFAs and pregnancy. Given the previous work in animal models demonstrating that relative concentrations of serum, PUFAs, and their metabolism are important to embryo implantation (9, 10), we also investigated relationships between n−6 to n−3 ratios (LA:ALA and AA:EPA) and IVF outcomes. Covariates associated with pregnancy were investigated further in a multivariable regression model controlling for patient age, BMI, history of prior pregnancy, male factor infertility, and diagnosis of endometriosis because these factors were associated with pregnancy in previous work from our center (6). A Spearman correlation coefficient was calculated by comparing relevant FFA values with embryo implantation rates (calculated as the number of gestational sacs on ultrasound/number of embryos transferred).

Results

Patient characteristics and IVF cycle outcomes

The characteristics of pregnant women were compared with those of women who did not get pregnant (Table 1). Patient ages ranged between 23 and 42 years. The groups did not differ by BMI, history of prior pregnancy, endometriosis, polycystic ovary syndrome, diminished ovarian reserve, tubal factor infertility, or unexplained infertility. However, pregnant women were younger (32.25 ± 4.36 vs 33.74 ± 4.45 years, P = .03), they had higher antral follicle counts (27.60 ± 16.09 vs 22.21 ± 11.13, P = .02), they were more likely to have a diagnosis of male factor infertility (relative risk 1.36, P = .02), and they had higher peak estradiol levels after controlled ovarian hyperstimulation (2716.66 ± 1166.81 vs 2327.17 ± 1062.13, P = .02).

Table 1.

Characteristics of Women Who Were Pregnant or Not Pregnant After Undergoing IVF

	Pregnant (n = 110)	Not Pregnant (n = 90)	RR (95% CI)	P Value
Patient characteristic
Age, y	32.6 ± 4.3	33.6 ± 4.6		.03
BMI, kg/m2	27.3 ± 6.8	28.1 ± 7.4		.4
Antral follicle count	27.6 ± 16.1	22.2 ± 11.1		.02
History of previous pregnancy	21	23	0.84 (0.6–1.2)	.3
Underwent IVF previously	8	7	1.2 (0.7–1.9)	.6
History of endometriosis	11	12	0.9 (0.6–1.3)	.5
Diagnosis of polycystic ovary syndrome	12	9	1.04 (0.7–1.6)	1.0
Male factor infertility	43	21	1.36 (1.1–1.7)	.02
Diminished ovarian reserve	5	10	0.59 (0.3–1.2)	.1
Tubal factor infertility	21	27	0.8 (0.5–1.1)	1.0
Unexplained infertility	18	8	1.3 (0.98–1.8)	.1
IVF cycle outcomes
Peak estradiol level	2716.7 ± 1166.8	2327.2 ± 1062.1		.02
No. of embryos transferred	2.26 ± 0.7	2.30 ± 0.8		.7
Blastocyst transfer	36/110	20/90	1.25 (0.97–1.61)	.1
Live birth	95/110

Abbreviations: CI, confidence interval; RR, relative rate.

Associations between serum PUFAs and pregnancy

No specific serum PUFAs were associated with pregnancy, but the LA:ALA ratios were. We divided women into tertiles based on the LA:ALA ratios and found a dose-dependent increase in the chance of pregnancy and the LA:ALA ratio. This result held after adjustment for patient age, antral follicle count, BMI, prior pregnancy, endometriosis, male factor infertility, and peak estradiol level (Table 2). After adjustment for relevant covariates, there was also a significant association between the LA:ALA ratio and the chance of a live birth (Table 2). We also found a correlation between LA:ALA ratios and embryo implantation rates (r = 0.21, P = .003). LA:ALA ratios were associated with AA:EPA ratios (r = 0.4, P < .001), another n−6 to n−3 ratio.

Table 2.

Results of Multivariable Logistic Regression Analysis Comparing LA:ALA Ratios in Women Who Had Clinical Pregnancies and Live Births After IVF vs Those Who Did Not

LA:ALA Tertile	RR (95% CI)	AOR (95% CI)a	Adjusted P Valuea
Clinical pregnancy after IVF
Tertile 1	Reference	Reference
Tertile 2	1.36 (0.96–1.94)	1.8 (0.86–3.8)	.1
Tertile 3	1.52 (1.09–2.13)	2.7 (1.3–5.8)	.009
Live birth after IVF
Tertile 1	Reference	Reference
Tertile 2	1.50 (1.01–2.23)	2.11 (0.98–4.56)	.06
Tertile 3	1.47 (0.99–2.18)	2.2 (1.04–4.76)	.04

Abbreviations: AOR, adjusted odds ratio; CI, confidence interval; RR, relative risk.

^aModel adjusted for BMI, age, antral follicle count, endometriosis history, history of prior pregnancy, history of male factor infertility, and peak estradiol level.

To determine whether the LA:ALA ratios were associated with embryo quality, serum LA:ALA ratios were compared between women who underwent cleavage (day 3) vs blastocyst stage transfer (day 5) because our embryo transfer protocols are designed to push women with better quality embryos to a blastocyst transfer. No differences were noted (14.0 ± 2.3 vs 14.5 ± 2.6, P = .2).

Discussion

In this study, increased serum LA:ALA ratios were associated with significantly increased odds of pregnancy and were also positively correlated with embryo implantation. Important opposing effects of n−6 and n−3 fatty acids and their derivatives have been shown in other contexts including cardiovascular function, platelet activation, and chemotaxis (11, 12). Our data suggest that the balance of the n−6 to n−3 PUFAs also plays a role in embryo implantation and pregnancy.

Knowledge of PUFA metabolism is helpful in understanding the relevance of n−6 to n−3 PUFA ratios to human health. In brief, n−6 and n−3 PUFAs compete for elongase and desaturase enzymes for conversion to longer-chain PUFAs. LA (n−6) is converted to γ-linoleic acid, dihomo-γ-linoleic acid, and then arachidonic acid (AA). ALA (n−3) is converted to stearidonic acid, eicosatetraenoic acid, and then eicosapentaenoic acid (EPA) and eventually docosahexanoic acid. AA and EPA can be mobilized from their cell membrane stores by phospholipase-A₂ and converted to eicosanoids by cyclooxygenase enzymes. Whereas AA is metabolized to the 2-series eicosanoids, such as prostaglandins (PGs) E₂ and F_2α, EPA is converted to the 3-series eicosanoids.

When metabolism of PUFAs is taken into consideration, there are several ways in which PUFAs may contribute to successful embryo implantation. First, eicosanoids derived from AA are generally proinflammatory, whereas those derived from EPA are largely anti-inflammatory (12). Inflammation is thought to play a key role at the time of embryo implantation (13). Therefore, an increase in the n−6 to n−3 balance at the time of implantation may enhance endometrial inflammation and encourage embryo implantation. PGs also play crucial roles in successful embryo implantation (14, 15). It is known that AA has a much stronger affinity for the prostaglandin endoperoxide synthase enzymes (1); therefore, the ability of EPA to produce PGs is much lower than that off AA. It is plausible that increased n−6 to n−3 PUFA ratio levels could lead to increased total PG synthesis, thus increasing the endometrial receptivity and embryo implantation. Alternatively, lower circulating n−6 to n−3 ratios may adversely affect oocyte quality and subsequent embryo development as suggested in recent work by Wakefield and colleagues (16). In a murine model, Wakefield and colleagues found that high ω-3 intake resulted in altered oocyte mitochondrial distribution and calcium levels and increased reactive oxygen species production. Embryonic developmental capacity was also impaired. In either case, whether PUFA intake affects endometrial receptivity or it affects the oocyte and subsequent embryo development, periconceptional PUFA supplementation warrants further study in women before clinical recommendations can be made.

We found a trend toward higher AA:EPA ratios in the women who became pregnant vs those who did not. Although this difference was not statistically significant, there was a correlation between LA:ALA and AA:EPA ratios. We suspect that the observed increase in the LA:ALA ratio leads to an increased AA:EPA ratio as well, but larger sample sizes may be needed to detect this difference. Alternatively, it is possible that a more dramatic difference in the ratio of long-chain PUFAs is evident in the endometrium itself, because PUFAs are primarily stored in an esterified form in the cell membranes. There are technical and ethical challenges to studying human endometrium at the time of implantation, which is why serum was an ideal surrogate in this study. Future studies using PUFA measurements in endometrial tissue from animal models and women using contraception are necessary to confirm and expand on our findings.

Dietary differences in the ratio of n−6 to n−3 PUFA intake are known to have functional effects. Specifically, diets enriched in n−3 PUFAs alter platelet function, blood pressure, and myocardial contractility (4, 11, 17). A diet high in linoleic acid fed to pregnant ewes led to increased PG synthesis in the endometrium (18). This study also concluded that the PG increase was a result of increased precursor availability rather than up-regulation of cyclooxygenase enzymes. Another study showed that diets high in n−3 PUFAs decreased the n−6:n−3 fatty acid ratio in the bovine uterus and also altered expression of genes involved in PG synthesis (19). Given these data and our current study, it is plausible that dietary manipulation through food or vitamin supplements could alter endometrial receptivity and subsequent embryo implantation rates.

In conclusion, we have shown that higher serum LA:ALA ratios are associated with an increased chance of pregnancy in women undergoing IVF. These data suggest that alterations of PUFAs, whether by diet or metabolism, may improve implantation rates during IVF. Whether serum PUFAs can be used as markers of uterine receptivity to determine timing and the number of embryos transferred deserves further study. Further investigation is also needed to determine whether serum PUFA levels correlate with endometrial PUFA content and to determine whether PUFAs affect human endometrial PG synthesis and subsequent correlation with establishment of pregnancy.