DHA Supplementation During Pregnancy Enhances Maternal Vagally Mediated Cardiac Autonomic Control in Humans

ABSTRACT

Background

DHA is an essential omega-3 (ω-3; n–3) fatty acid that has well-established benefits for the fetus. DHA also has the potential to influence the health of the mother, but this area is understudied.

Objectives

The objective of this secondary analysis was to determine if DHA was related to maternal heart rate (HR) and heart rate variability (HRV) metrics in a large cohort of pregnant women.

Methods

In the parent trial (1R01HD086001) eligible participants (≥18 y old, English speaking, carrying a singleton pregnancy, 12–20 wk of gestation) were randomly assigned to consume 200 mg/d or 800 mg/d DHA for the duration of their pregnancy (n = 300). Weight, blood pressure, and magnetocardiograms (MCGs) were collected at 32 wk and 36 wk of gestation (n = 221). Measures of HR and HRV in time-, frequency-, and nonlinear-domains were determined from the isolated maternal MCG. Treatment group and timepoint were examined as predictors in association with HR and HRV metrics using random-intercept mixed-effects ANOVA unadjusted and adjusted models accounting for weight and dietary DHA intake.

Results

Women receiving the higher dose of DHA (800 mg/d) during pregnancy had lower HR, lower sympathetic index, higher vagally mediated HRV indices, and greater HRV complexity when compared with the women who received the lower dose (200 mg/d; all P < 0.05). All the dose relations remained significant even after controlling for the effect of time, maternal weight, and dietary DHA intake.

Conclusions

DHA supplementation increases vagal tone in pregnant women. Longitudinal studies examining the potential link between DHA, enhanced vagal tone, and reported reduction in early preterm birth are warranted.

Introduction

Maternal health is declining in the United States and rates of maternal mortality are the highest of any other developed nation (1). The prevalences of chronic conditions such as hypertension, type 2 diabetes, and obesity increased by 31–100% respectively in commercially insured patients from 2014 to 2018 (2). Concurrently, rates of pregnancy complications such as gestational diabetes and pre-eclampsia, and childbirth complications such as eclampsia, cardiomyopathy, and embolism have increased (2). The CDC and American College of Gynecology agree that improving maternal health should be a national priority; however, few studies have identified effective strategies to alleviate the burden of poor health in this population.

In healthy women, pregnancy results in significant maternal cardiovascular and metabolic adaptations. As blood volume increases, heart rate (HR) and cardiac output increase, and heart rate variability (HRV) decreases (3). The autonomic nervous system (ANS) plays an important regulatory role as the body adapts to these normal physiological changes across the course of gestation (4, 5). Longitudinal measures of maternal ANS function in healthy pregnancies describe an increase in vagal activity and reduced sympathetic activity in the first trimester (6) with a shift toward reduced vagal and increased sympathetic activity in the last 2 trimesters of pregnancy (7, 8). In women who have chronic health conditions or pregnancy complications, the shift in sympathetic activation is likely exacerbated, although this area is understudied.

DHA (22:6n–3) is a long-chain PUFA found in all cell membranes. Meta-analyses of randomized controlled trials in nonpregnant adult humans show fish oil supplements containing both DHA and EPA (20:5n–3) lower HR (9, 10) and increase HRV, attributed to increased vagal tone (11). The trials of DHA supplementation in pregnancy have focused primarily on the effect of maternal supplementation on birth outcomes and offspring health. Higher-dose (1000 mg/d) maternal supplementation was found to significantly reduce the risk of early preterm birth in women with low DHA status—defined as <6 weight percentage total fatty acids of RBC phospholipids (12). When pregnant women were randomly assigned to receive 600 mg/d DHA compared with placebo during the last 2 trimesters, offspring of those receiving DHA were shown to have a higher fetal autonomic brain age score (13), increased fetal HRV, and greater newborn neurobehavior in autonomic and motor clusters of the Neonatal Behavioral Assessment Scale (14). Despite the known effects of long-chain PUFAs on adult and newborn cardiac autonomic control, there are no randomized controlled trials that have reported the effects of supplemental DHA consumed during pregnancy on maternal HR and HRV.

The goal of this secondary analysis was to determine the effect of DHA on maternal HR and HRV at 32 and 36 wk of gestation. We hypothesized that women randomly assigned to receive a higher dose (800 mg/d DHA) would have lower HR and higher HRV than women who received a lower dose commonly found in prenatal vitamins (200 mg/d). The analyses presented here will use data from the Prenatal Autonomic Neurodevelopmental Assessment (PANDA; NCT02709239), a recently completed double-blind randomized controlled clinical trial in which pregnant women were randomly assigned to receive either 200 or 800 mg supplemental DHA from enrollment at 12–20 wk gestational age to delivery. The PANDA trial was designed to evaluate the effect of DHA dose on the incidence of maternal-fetal DHA RBC equilibrium (a potential index of sufficient maternal DHA for fetal transfer) and the effect of equilibrium and DHA status on fetal HRV and neurodevelopment (15). Because biomagnetometry was used to record simultaneous maternal and fetal magnetocardiograms (MCGs), we were also able to isolate maternal MCGs and calculate HR and HRV metrics.

Methods

Trial design

Methodological details have been reported elsewhere (15). Briefly, the parent trial was a randomized, longitudinal, double-blind, single-center, phase III superiority trial. The pregnancy phase of the trial was conducted at the University of Kansas Medical Center, USA between June 2016 and September 2020, approved by the Human Subjects Committee (STUDY 00003792) in accordance with the Helsinki Declaration of 1975 as revised in 1983. The trial was overseen and annually reviewed by a Data Safety and Monitoring Board. The trial is registered at https://www.clinicaltrials.gov/ct2/show/NCT02709239.

Trial participants

Pregnant women were screened for eligibility between June 2016 and March 2020. To be eligible for the study, women had to be ≥18 y old, English speaking, carrying a singleton pregnancy, between 12 and 20 wk of gestation, and not having had a prepregnancy BMI ≤18.5 kg/m², or body weight >250 pounds (113 kg) at enrollment. The latter exclusion was due to the safety limit of the biomagnetometer support chair. Women were excluded from participation if they had a known serious illness likely to result in hospitalization or threaten pregnancy, for example, cancer, lupus, hepatitis, type 1 diabetes, hypertension, or self-reported drug or alcohol abuse. Following screening and written informed consent, 300 women were randomly assigned to the capsule allocation (Figure 1).

FIGURE 1.

Participant flow. MCG, magnetocardiocram.

Randomization and blinding

Women were randomly assigned to receive 4 capsules containing algal oil that provided a total of either 200 mg/d (50 mg/capsule) or 800 mg/d (200 mg/capsule) DHA daily (Life's DHA-S oil; DSM Nutritional Products LLC). Capsules were taken with or without food. The computer-generated randomization schedule was provided by the study statistician to the Investigational Pharmacy at the University of Kansas Health System. Throughout the duration of the study, only the pharmacy knew the subject allocation; study team members and participants were blinded to the capsule assignment. Capsules of the same color and size were packaged identically in opaque bottles to prevent study personnel and participants from guessing their group allocation. Participants received their first bottle from study team members upon enrollment and thereafter received capsules every 30 d, mailed from the Investigational Pharmacy with instructions to return any unused capsules from the previous 30 d in a return mailer. Women were instructed to stop taking capsules at delivery. Participants were contacted monthly until delivery. Adverse events were recorded and have been reported in detail, and no major safety concerns were observed in either group (15).

Maternal characteristics

Maternal age was calculated at the date of enrollment using date of birth. Maternal prepregnancy BMI was calculated based on self-reported prepregnancy weight and height at the time of screening. Maternal weight was recorded at 32 and 36 wk of gestation on a calibrated scale. Participants were asked to remove shoes and extra clothing (e.g., sweaters and jackets) prior to the weight measurement. Maternal blood pressure (systolic and diastolic) was also measured at 32 and 36 wk of gestation with a digital blood pressure device and a cuff appropriate to the mother's arm size. Maternal dietary DHA intake was captured using the Diet History Questionnaire-II at 32 wk of gestation using the version reporting past month's intake with portion size (16), and the Diet*Calc software (17) was used to calculate DHA intake in grams/day.

Magnetocardiogram

Magnetocardiograms (MCGs) are a noninvasive technique to measure the magnetic fields naturally generated from the heart, similar to an electrocardiogram. Simultaneous maternal-fetal MCGs were recorded at 32 and 36 wk of gestation using an 83-channel dedicated fetal biomagnetometer (CTF MEG) in a magnetically shielded room. Women were seated in a slightly reclining position with the gravid abdomen making slight contact with the biomagnetometer sensor array. MCGs were recorded for 30 min using a 1200-Hz sampling rate with a recording filter of 0–75 Hz. Raw, deidentified data were stored on secure University servers.

After the recording was complete, a bidirectional fourth-order Butterworth filter (2–40 Hz) was applied to digitally filter the raw data for offline processing. To separate maternal MCG from fetal MCG, the data were presented to an Infomax Independent Components Analysis (ICA) algorithm in EEGLAB toolbox (version 4.311) (18). Using this method, the spatially distinct maternal ICA components derived from the maternal MCG were identified and summed to reconstruct the signal in channel space.

HRV analysis

After reconstructing the maternal MCG, single-channel data with the most typical QRS complex was exported as an ASCII text file and then imported into Kubios HRV software (version 3.4.2) (19). All R-peaks were marked and ectopic beats corrected using an automatic artifact correction algorithm and then visually inspected for errors or missed beats. Kubios outputs multiple HRV metrics in time-, frequency-, and nonlinear-domains. We used measures of HRV across domains for a comprehensive picture of autonomic modulation.

In the time-domain, we measured HR in beats per minute (bpm), and deceleration capacity (DC, milliseconds), a metric that quantifies heart rate decelerations, considered an index of vagal modulation.

In the frequency-domain, we quantified power in the very-low-frequency (VLF; 0.0–0.04 Hz), low-frequency (LF; 0.04–0.15 Hz), and high-frequency (HF; 0.15–0.40 Hz) bands. These measures represent the power spectral density of the RR interval series using an autoregressive modeling method. VLF power is influenced by many physiological mechanisms including the sympathetic system, which contributes to the amplitude and frequency within this band (20). LF power is generally considered an index of overall HRV with both sympathetic and vagal contributions, whereas HF power is an index of vagal function linked to respiration (respiratory sinus arrhythmia).

In the nonlinear-domain we obtained 3 indices from the Poincaré plot: 1) SD1, the standard deviation of the short-term normal-to-normal RR intervals. SD1 is a vagally mediated metric and is perfectly correlated (R² = 1.0) with the time-domain measure, RMSSD (root mean square of successive differences in milliseconds). 2) SD2, the standard deviation of the long-term normal-to-normal RR intervals, an index of overall HRV, influenced by both vagal and sympathetic activity. SD2 is highly correlated (R² = 1.0) with the time-domain metric, SDNN (standard deviation of normal-to-normal RR intervals in milliseconds) (21). 3) SD2:SD1, a ratio representing nonlinear characteristics of HRV, highly correlated with metrics related to fractal scaling and entropy (21). Additionally, we measured sample entropy, an index of signal complexity or irregularity, and DFA1 (detrended fluctuation α-1), an index of short-term fluctuations (n = 4–12 beats) in a signal that represents the correlation within a time-series. An increase in sample entropy indicates a more complex and variable time series. Conversely, an increase in DFA1 would indicate that the time series is more highly correlated and therefore less complex. We used 2 comprehensive indices: 1) the parasympathetic nervous system (PNS) index, calculated using the mean RR interval, the time-domain measure RMSSD, and SD1 in normalized units; and 2) the sympathetic nervous system (SNS) index, calculated using mean HR, Baevsky stress index (22), and SD2.

Statistical analysis

Women were included in statistical analyses according to the treatment group for which they were randomly assigned at parent study enrollment under intent-to-treat principles. The distributions of all continuous variables were examined for normality. Descriptive statistics were calculated, including means and SDs for normally distributed variables, and medians and IQRs for skewed variables. Log transformations were applied to skewed maternal HRV variables prior to analysis. Mean values for maternal HRV variables by treatment group and visit (32 and 36 wk) were examined visually using line plots, and SEs were corrected for within-subjects correlation by visit. Paired t-tests compared mean maternal heart rate, blood pressure, and HRV metrics at 32 and 36 wk. Treatment group and time of visit (32 and 36 wk) were examined as predictors in association with maternal heart rate, blood pressure, and HRV metrics using random-intercept mixed effects ANOVA models with an unstructured covariance matrix and Kenward–Roger approximation for Df. Interactions between group and time were tested for all models, though none were statistically significant nor retained in final models. Models were also adjusted for maternal weight at 32 and 36 wk as well as dietary DHA self-reported at 32 wk. Statistical significance for all analyses was defined as P < 0.05, and statistical analyses were performed using SAS software, version 9.4 (SAS Institute, Inc) or R, version 4.1.1 (R Foundation).

Results

Three hundred participants enrolled in the parent trial; 214 women had MCG data at 32 and 36 wk of gestation and were included in this analysis. A complete CONSORT diagram including reasons for withdrawal or removal was previously published (15) and a participant flow for this secondary analysis is shown in Figure 1. Relevant maternal characteristics, DHA status, and HRV metrics by group and timepoint are shown in Table 1. Rates of adherence to capsule consumption were similar: 24.0 capsules/wk for the 200-mg/d group and 23.7 capsules/wk for the 800-mg/d group.

TABLE 1.

Characteristics of pregnant women who received 200 or 800 mg DHA/d from 12 to 20 wk of gestation through delivery¹

	200 mg/d Mean ± SD n = 114	800 mg/d Mean ± SD n = 107
Maternal age at enrollment, y	30.7 ± 5.18	30.2 ± 4.75
Prepregnancy BMI, kg/m2	27.2 ± 5.05	26.3 ± 5.89
Dietary DHA intake in pregnancy, mg/d	60.9 ± 66.6	55.5 ± 64.4
	32 wk		36 wk
	200 mg/d Mean ± SD n = 114	800 mg/d Mean ± SD n = 107	200 mg/d Mean ± SD n = 114	800 mg/d Mean ± SD n = 107
Maternal weight, kg	85.9 ± 13.6	82.5 ± 15.1	88.4 ± 13.9	85.1 ± 15.4
Blood pressure (systolic),3 mmHg	115 ± 11.9	115.1 ± 9.37	123 ± 13.6	121 ± 12.4
Blood pressure (diastolic),3 mmHg	70.8 ± 8.64	70.0 ± 8.02	74.0 ± 9.53	73.2 ± 7.86
HR, bpm	85.7 ± 10.4	83.0 ± 11.9	87.7 ± 11.7	84.1 ± 12.4
PNS index	−1.53 ± 0.885	−1.36 ± 0.902	−1.72 ± 0.882	−1.41 ± 0.967
SNS index	2.78 ± 1.69	2.36 ± 1.79	3.12 ± 1.94	2.55 ± 2.01
DC,2 ms	24.5 (15.1, 37.4)	27.0 (16.8, 41.3)	20.0 (12.1, 32.2)	27.0 (17.3, 39.1)
log DC, ms	3.11 ± 0.664	3.26 ± 0.657	2.95 ± 0.720	3.25 ± 0.683
SD2%2	62.9 (50.1, 81.6)	63.5 (51.2, 84.7)	64.2 (51.8, 94.2)	73.7 (58.9, 88.7)
log SD2%	4.16 ± 0.403	4.17 ± 0.335	4.21 ± 0.409	4.26 ± 0.352
SD1%2	15.2 (9.64, 23.3)	17.0 (12.3, 24.9)	12.3 (9.05, 20.9)	16.4 (11.8, 24.8)
log SD1%	2.72 ± 0.654	2.83 ± 0.564	2.59 ± 0.647	2.81 ± 0.623
SD2:SD1	4.79 ± 2.52	4.25 ± 2.19	5.75 ± 3.33	4.78 ± 2.63
log VLF power AR	7.19 ± 0.839	7.19 ± 0.686	7.39 ± 0.830	7.46 ± 0.721
log LF power AR	5.56 ± 0.998	5.61 ± 0.904	5.37 ± 0.979	5.56 ± 0.856
log HF power AR	5.57 ± 1.20	5.76 ± 1.14	5.33 ± 1.22	5.77 ± 1.22
DFA1	1.10 ± 0.278	1.04 ± 0.287	1.14 ± 0.269	1.05 ± 0.272
Sample entropy	1.17 ± 0.316	1.27 ± 0.319	1.06 ± 0.359	1.16 ± 0.313

¹AR, autoregressive; DC, deceleration capacity; DFA1, detrended fluctuation α-1; HF, high-frequency; HR, heart rate; LF, low-frequency; PNS, parasympathetic nervous system; SD1, standard deviation of the short-term normal-to-normal RR intervals; SD2, standard deviation of the long-term normal-to-normal RR intervals; SNS, sympathetic nervous system; VLF, very-low-frequency.

²Median and IQR displayed.

³Data missing for 1 participant in the 200-mg/d group and 2 participants in the 800-mg/d group at 36 wk.

Results from the unadjusted and adjusted mixed effects models are shown in Table 2. After adjusting for maternal weight and dietary DHA, the DHA dose had a significant effect on several key HRV metrics. Women in the 800-mg/d group showed significantly lower maternal HR (β = −3.2; P = 0.02), SNS index scores (β = −0.5; P = 0.03), and SD2:SD1 ratio (β = −0.8; P = 0.01) when compared with the 200-mg/d group, suggesting that higher DHA supplementation reduced sympathetic activity. All vagally mediated metrics were significantly higher in women in the 800-mg/d group when compared with the 200-mg/d group in adjusted models: PNS index (β = 0.2; P = 0.02), log DC (β = 0.2; P = 0.01), log SD1 (β = 0.2; P = 0.02), and HF power (β = 0.3; P = 0.03).

TABLE 2.

Relation between treatment group, timepoint, and maternal HRV metrics (n = 214)¹

	Unadjusted				Adjusted for maternal weight and dietary DHA
	Treatment (800 vs. 200 mg/d)		Visit (36 vs. 32 wk)		Treatment (800 vs. 200 mg/d)		Visit (36 vs. 32 wk)
	β (SE)	P value	β (SE)	P value	β (SE)	P value	β (SE)	P value
HR, bpm	−3.81 (1.41)	0.007	1.61 (0.575)	0.006	−3.21 (1.41)	0.024	1.27 (0.600)	0.036
BP (systolic), mmHg	−0.080 (1.26)	0.945	6.24 (0.817)	<0.0001	0.681 (1.138)	0.551	5.42 (0.827)	<0.0001
BP (diastolic), mmHg	−0.345 (0.948)	0.716	3.13 (0.501)	<0.0001	−0.139 (0.908)	0.879	2.79 (0.509)	<0.0001
Sympathetic index and metrics of overall HRV (both sympathetic and vagal contributions)
SNS index	−0.585 (0.222)	0.009	0.271 (0.101)	0.008	−0.492 (0.224)	0.029	0.217 (0.105)	0.040
SD2:SD1	−0.830 (0.298)	0.006	0.751 (0.171)	<0.0001	−0.800 (0.299)	0.008	0.777 (0.177)	<0.0001
log SD2	0.0198 (0.0447)	0.658	0.074 (0.021)	0.0006	0.000460 (0.0450)	0.992	0.0893 (0.0219)	<0.0001
VLF power	0.0175 (0.0914)	0.849	0.229 (0.0442)	<0.0001	−0.0177 (0.0923)	0.849	0.260 (0.0454)	<0.0001
LF power	0.133 (0.110)	0.229	−0.121 (0.0553)	0.030	0.0820 (0.113)	0.467	−0.103 (0.0573)	0.075
Vagal metrics
PNS index	0.292 (0.107)	0.007	−0.130 (0.0565)	0.022	0.259 (0.108)	0.017	−0.109 (0.0587)	0.066
log DC	0.231 (0.0811)	0.005	−0.0885 (0.0363)	0.016	0.210 (0.0821)	0.011	−0.0701 (0.0379)	0.066
log SD1	0.189 (0.0723)	0.009	−0.0716 (0.0386)	0.065	0.170 (0.0736)	0.022	−0.0596 (0.0401)	0.139
HF power	0.352 (0.141)	0.013	−0.120 (0.0695)	0.085	0.313 (0.143)	0.029	−0.0901 (0.0724)	0.214
Complexity metrics
SampleEn	0.124 (0.0373)	0.001	−0.112 (0.0212)	<0.0001	0.119 (0.0380)	0.002	−0.116 (0.0217)	<0.0001
DFA1	−0.0955 (0.0318)	0.003	0.0239 (0.0170)	0.161	−0.0968 (0.0322)	0.003	0.0214 (0.0175)	0.224

¹Coefficients and P values correspond to fixed effects estimates. Excludes n = 7 women for missing covariate data. BP, blood pressure; DC, deceleration capacity; DFA1, detrended fluctuation α-1; HF, high-frequency; HR, heart rate; HRV, heart rate variability; LF, low-frequency; PNS, parasympathetic nervous system; SampleEn, sample entropy; SD1, standard deviation of the short-term normal-to-normal RR intervals; SD2, standard deviation of the long-term normal-to-normal RR intervals; SNS, sympathetic nervous system; VLF, very-low-frequency.

The complexity measures showed similar associations with DHA treatment between unadjusted and adjusted models (Table 2). Sample entropy was significantly higher in the 800-mg/d group (adjusted β = 0.1; P = 0.002), whereas DFA1 was lower in the 800-mg/d group (β = −0.1; P = 0.003), indicating weaker correlations and, thereby, more complexity in the HR time series. There was no treatment group difference for maternal blood pressure, SD2, VLF, or LF power.

As expected, there was a significant increase in maternal blood pressure between the 2 visits that remained statistically significant after adjusting for maternal weight and dietary DHA intake (systolic: β = 5.2; P < 0.0001; diastolic: β = 2.6; P < 0.0001; Table 2). Maternal HR increased at 36 wk compared with 32 wk (β = 1.3; P = 0.04), consistent with an increase in the SNS index between the 2 visits (β = 0.2; P = 0.04). Significant increases over time were also observed in the frequency domain metric, VLF power (β = 0.3; P < 0.0001), log SD2, the nonlinear index of overall HRV (β = 0.1; P < 0.0001), and the ratio of SD2:SD1 (β = 0.8; P < 0.0001) in adjusted models. LF power significantly decreased over time in the unadjusted model (β = −0.1; P = 0.03), though this effect was attenuated after adjusting for maternal weight and dietary DHA intake (β = −0.1; P = 0.08). Results of the vagally mediated metrics were mixed. Unadjusted models showed significant decreases in the PNS index (β = −0.1; P = 0.02) and DC (β = −0.09; P = 0.02) over time, which reached marginal significance in adjusted models (PNS index: β = −0.1; P = 0.02; DC: β = −0.07; P = 0.07). There was no significant change in HF power or SD1 between the 2 visits. For measures indicative of HR complexity, sample entropy was strongly reduced over time (β = −0.1; P < 0.0001). Time was not associated with DFA1. These results are consistent with the reported shift towards higher sympathetic and reduced vagal activity along with a reduction in HR complexity, a reflection of the inability to rapidly vary HR to physiological demand with advancing gestation.

Of note, in adjusted models maternal weight was significantly associated with HR (β = 0.2; P = 0.001), blood pressure (systolic: β = 0.3; P < 0.0001; diastolic: β = 0.2; P < 0.0001), SNS index (β = 0.02; P = 0.002), SD2 (β = −0.005; P = 0.0008), VLF power (β = −0.01; P = 0.002), PNS index (β = −0.01; P = 0.01), DC (β = −0.007; P = 0.01), SD1 (β = −0.006; P = 0.03), and HF power (β = −0.01; P = 0.01). Interestingly, maternal weight was not a predictor of HR complexity as indexed by sample entropy and DFA1.

Discussion

The goal of this secondary analysis was to determine the effect of DHA on maternal physiology (e.g., HR, HRV) during a randomized controlled trial of DHA supplementation (200 mg/d compared with 800 mg/d) during pregnancy. Women who received the higher dose of DHA during pregnancy had improved autonomic physiology compared with the women who received the lower dose. There was no relation between dose and maternal blood pressure. All the dose relations remained significant when accounting for the effect of time, maternal weight, and dietary DHA intake. These novel findings suggest a potential benefit of DHA for pregnant women, which could be especially relevant for those with risk factors that are associated with increased sympathetic activity, such as maternal obesity (23), gestational diabetes (24), and pre-eclampsia (25). Autonomic dysregulation, characterized as a trend toward elevated sympathetic tone and lower vagal tone, has been implicated in pre-eclampsia, a hypertensive disorder of pregnancy (26). The precise mechanism leading to the development of pre-eclampsia is not fully understood. Autonomic dysregulation might play a role in the pathophysiology leading to pre-eclampsia, but it is worth noting that there was no DHA dose effect for maternal blood pressure in the current study. There is some evidence that pre-eclampsia is associated with lower maternal concentrations of both n–6 and n–3 long-chain PUFAs compared with healthy pregnancies (27).

Due to the considerable changes in maternal cardiovascular and autonomic systems as gestation advances, it was necessary to consider the effect of increasing gestation in our analyses. From 32 to 36 wk of gestation, maternal HR and blood pressure increased whereas vagally mediated metrics of HRV decreased. These significant changes over time support current evidence that the physiological demands of pregnancy result in a shift toward greater sympathetic dominance, reduced vagal activity, and less HR complexity in the third trimester (6, 28). Women in both the 800-mg/d and 200-mg/d groups experienced the same shift toward greater sympathetic dominance across gestation, but the effect was diminished in the 800-mg/d group, indicated by lower HR, higher vagally mediated HRV, and greater HR complexity (Figure 2). However, no suggestion of significant statistical interaction was observed between DHA treatment dose and time for these metrics.

FIGURE 2.

Mean HR and HRV metrics by timepoint and treatment group. Values represent means by timepoint and treatment group ± SEMs, corrected for within-subjects correlation by visit as described in Morey (51). n = 221. A. HR, heart rate; B. SD1%, standard deviation of the short-term normal-to-normal RR intervals (vagally mediated); C. SD2% standard deviation of the long-term normal-to-normal RR intervals (overall HRV); D. SD2:SD1: a ratio that represents nonlinear characteristics of HR; E Sample entropy, an index of signal complexity or irregularity.

In the adjusted models, maternal weight was significantly and independently associated with nearly every outcome. Specifically, increased maternal weight predicted higher maternal HR, blood pressure, and SNS index along with reduced overall and vagally mediated metrics of HRV. Interestingly, there was no relation between maternal weight and the complexity measures, sample entropy and DFA1. This suggests that weight has no relation with mechanisms that control the ability to rapidly vary both the rate and magnitude of change in HR beyond those changes associated with the physiological demand of pregnancy.

DHA is known to influence HR and HRV in the fetus (14, 29), infants (30, 31), children, and nonpregnant adults (32). To our knowledge, this is the first report of a similar effect in pregnant women. The precise mechanism of the effect of DHA on cardiac autonomic regulation is not fully elucidated and beyond the scope of this report; however, a thorough review of potential mechanisms was recently published (32). DHA is incorporated into cardiac myocytes (33) and some results suggest a direct effect on the myocardium in transplant patients with vagal denervation (34). In post–myocardial infarct patients, supplemental DHA resulted in a significant increase in vagally mediated HF power, with no change in metrics of overall HRV (35), identical to what we report here. Because DHA is incorporated into all cell membranes, it is unlikely that a direct effect on cardiac myocytes is the only mechanism of action that leads to enhanced vagal function. Preganglionic parasympathetic neurons are found in the brainstem nucleus ambiguus. A direct injection of DHA into the nucleus ambiguus of rats resulted in lower HR of ∼25 bpm and lasted >30 min (36).

The vagus nerve traverses the body and has emerged as a key regulator of inflammation via cholinergic modulation, as reviewed by Alen (37). Experimental results of direct vagus nerve stimulation on inflammatory markers have been mixed. However, electrical stimulation of the human vagus has been shown to produce endogenous specialized proresolving lipid mediators (SPMs) that are synthesized from long-chain n–3 fatty acids, DHA and EPA, and n–6 arachidonic acid (38). DHA plays an important role in the regulation of inflammatory processes, but also serves as a substrate for proresolving D-series resolvins, protectins, and maresins (39). A recent study of 293 pregnant women found higher resolvin concentrations in women with pre-eclampsia and metabolic syndrome, with a significant positive association with blood pressure (40), not unexpected as both conditions are chronic inflammatory states. Supplementing pregnant women with fish oil–derived n–3 fatty acids increased the placental accumulation of SPM precursors (41). There is a considerable knowledge gap surrounding the role of SPMs in pregnancy and placental function. Our understanding of the relation between vagal activity, n–3 PUFAs, and SPMs is limited and deserves further research.

Perhaps one of the most important findings related to DHA supplementation in pregnancy is the reduction in early preterm birth (42–44), particularly in women with low DHA status (12). Although the precise mechanism is unknown, it is feasible that it is due in part to the anti-inflammatory effects of enhanced vagal function. Further, higher circulating DHA would provide the substrate for the synthesis of SPMs. Therefore, the proresolving effects of increased concentrations of SPMs could contribute to the reduction in early preterm birth observed in these clinical trials. Indeed, secondary analysis of Carlson et. al. (45) found higher-dose DHA modulated the immune response during pregnancy and this was associated with a decrease in preterm birth.

These studies also underscore the importance of knowing maternal DHA status and supplementing accordingly. Recently, we published results showing that using a 7-question screener questionnaire was highly correlated with RBC DHA in 1355 pregnant women (46). Of note, maternal age, parity, and socioeconomic status were significant predictors of maternal DHA status. In this sample, the average intake of DHA (∼60 mg/d) from food and supplements was low.

Maternal morbidity and mortality continue to increase in the United States along with greater health disparities for women of color (47). One concept thought to explain the physiological effects over the life-course is allostatic load, defined as the chronic exposure to an external stressor (poverty, disparities, poor nutrition) and the hyperactivation of hormonal, metabolic, and cardiovascular systems resulting in chronic disease (48). Because vagal tone is a regulator of allostatic systems, it is not surprising that there is good agreement between vagal measures of HRV and allostatic load assessments (49). Given the effect of DHA in lowering HR, enhancing vagal function, and reducing the rate of early preterm birth, we believe there is a direct and immediate benefit to recommending supplementation in a clinical setting in women with low dietary DHA intake, before conception and throughout pregnancy and lactation (50).

We were limited by the original study design where the primary outcome was fetal neurodevelopment at 32 and 36 wk. Ideally, future studies would include longitudinal measures from the first through third trimesters in both typical and complicated pregnancies. These measures should include comprehensive measures of HRV in time-, frequency-, and nonlinear-domains. Improving maternal health is an essential endeavor during a time when pregnant women are experiencing increasing rates of morbidity and mortality. We found that supplementing with 800 mg/d DHA during pregnancy lowers HR and improves metrics of vagally mediated HRV when compared with a dose typically found in prenatal vitamins (200 mg). These results increase our knowledge of the effects of DHA supplementation in pregnant women and could provide important clues to explain the link between DHA and the reduction of early preterm birth reported in several clinical trials.

Notes

This study was supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) 1R01HD086001 (US NIH Grant/Contract).

Author disclosures: SEC and JC have received honorariums for presentations about DHA in infancy and pregnancy. The other authors have no competing interests.

Abbreviations used: ANS, autonomic nervous system; bpm, beats per minute; DC, deceleration capacity; DFA1, detrended fluctuation α-1; HF, high-frequency; HR, heart rate; HRV, heart rate variability; ICA, Independent Components Analysis; LF, low-frequency; MCG, magnetocardiogram; PANDA, Prenatal Autonomic Neurodevelopmental Assessment; PNS, parasympathetic nervous system; SD1, standard deviation of the short-term normal-to-normal RR intervals; SD2, standard deviation of the long-term normal-to-normal RR intervals; SNS, sympathetic nervous system; SPM, specialized proresolving lipid mediator; VLF, very-low-frequency.

Data Availability

We are willing to share deidentified data from the study with a signed data access agreement that includes the study principal investigators, contingent on approval of the planned use of the data. Because the data are entered into an electronic system, a specific request to KMG (kgustafson@kumc.edu) or DNC (dchristifano@kumc.edu) would be needed to generate a data output for other investigators. We plan to publish secondary results and longitudinal outcomes of the trial and cannot share some data until the study is final.