Impact of low-level prenatal alcohol exposure and maternal stress on autonomic regulation

Abstract

BACKGROUND:

Prenatal alcohol exposure (PAE) impacts the neurodevelopment of the fetus, including the infant’s ability to self-regulate. Heart rate variability (HRV), that is, the beat-to-beat variability in heart rate, is a non-invasive measurement that can indicate autonomic nervous system (ANS) function/dysfunction.

METHODS:

The study consisted of a subset of our ENRICH-2 cohort: 80 participants (32 PAE and 48 Controls) who had completed three visits during pregnancy. The participants completed a comprehensive assessment of PAE and other substances throughout pregnancy and assessments for stress, anxiety, and depression in the third trimester. At 24 h of age, infant HRV was assessed in the hospital during the clinically indicated heel lance; 3- to 5-min HRV epochs were obtained during baseline, heel lancing, and recovery episodes.

RESULTS:

Parameters of HRV differed in infants with PAE compared to Controls during the recovery phase of the heel lance (respiratory sinus arrhythmia (RSA) and high-frequency (HF), p < 0.05). Increased maternal stress was also strongly associated with abnormalities in RSA, HF, and low-frequency / high-frequency (LF/HF, p’s < 0.05).

CONCLUSIONS:

Alterations in ANS regulation associated with PAE and maternal stress may reflect abnormal development of the hypothalamic-pituitary-adrenal axis and have long term implications for infant responsiveness and self-regulation.

INTRODUCTION

Alcohol use during pregnancy continues to pose a worldwide concern.^1–5 From 1984 to 2014, the prevalence of alcohol use during pregnancy was estimated at 9.8%, with women in the United States having a prevalence of alcohol use of 10.2%.² More current reports estimate 11.5–15.0% of pregnant women consume alcohol during pregnancy.^6–8 The recent COVID-19 pandemic has also resulted in an increase in alcohol consumption.⁹ Importantly, alcohol can readily cross the placenta¹⁰ and impact the developing fetus. As the fetus is not able to effectively eliminate alcohol, exposure can have a prolonged effect.^11–13 Therefore, the CDC and researchers believe that no or only a small amount of alcohol use during pregnancy is considered safe.¹⁴

The effect of alcohol on the developing fetus is broad, with multiple organ systems impacted, including cardiac, endocrine, and immunological.^2,10,15–20 Most notably, the nervous system is directly impacted by prenatal alcohol exposure (PAE).^21–25 Fetal alcohol spectrum disorder (FASD) is the umbrella term used to describe the long-term impact of PAE, the most serious of which is fetal alcohol syndrome (FAS). Individuals with an FASD experience difficulties in motor functioning, attention, executive function, cognition, and social skills, as well as exhibiting adverse physical and mental health outcomes.^{14,23,26–30}

One of the problems in providing services and support for children with FASD and their families is centered around the difficulty in obtaining a diagnosis at an early age, so that early intervention services can be initiated.^31–33 Often, the extent of the child’s deficits may not be apparent until middle childhood.³³ Thus, identifying markers or measurements that can provide support for a diagnosis is essential in this population and continues to be a focus of research.^31,34

Self-regulation, or the ability to monitor and manage feeling and behaviors, is critically important in the development and maturation of executive function and is currently being investigated as a possible early marker of altered development in numerous studies.^35,36 Indeed, updated guidelines on diagnosing an FASD include neurobehavioral impairment in specific domains, such as global intellectual ability, cognition, behavior and self-regulation, and/or adaptive skills.³² The behavioral deficits in children with a FASD may present more challenges in daily life than deficits in other domains.

Heart rate variability (HRV) can provide insight into the balance between the sympathetic and parasympathetic nervous system activity and ultimately into emerging self-regulatory skills. Specifically, the variability in heart rate, or the beat-to-beat differences, results from changing levels of sympathetic and parasympathetic outflow.^37,38 HRV signal parameters can be used to provide a noninvasive assessment of the autonomic nervous system.^37,38 Numerous medical conditions in infants, including prematurity, low birth weight, sepsis, necrotizing enterocolitis, and hypoxic ischemic encephalopathy, have been associated with decreased HRV; thus, measurement of HRV has been useful in indicating developmental alterations.^39–43 Interestingly, prior studies of effects of moderate to heavy PAE, such as ref. ⁴⁴ reported that the alcohol-exposed infants had decreased HRV in response to heel lancing for the newborn screening compared to unexposed healthy controls, indicative of reduced behavioral arousal and alteration in self-regulation. Additional studies have reported similarly decreased HRV in infants with PAE and reductions in parasympathetic activity during infancy that persist into childhood.^45,46 To date, no studies of low levels of PAE using HRV have been conducted, making the applicability of these results unknown in milder exposures.

While prenatal substance exposure can impact infant HRV parameters, another important contributing factor is maternal stress, which consists of perceived stress, anxiety, and associated mental health disorders. Numerous studies have investigated the impact of maternal stress on infant and child self-regulation.^47–50 Specifically, maternal stress has been shown to be directly associated with lower infant physiological regulation at 6 months of age,⁵⁰ with an increased risk of self-regulatory difficulties at age 2 years.⁴⁹ Thus, to better understand the impact of PAE on HRV, information on maternal stress also needs to be considered.

While most prior studies on effects of PAE have been conducted in populations with moderate-to-heavy alcohol use during pregnancy, little is known about the effects of lower levels of PAE on infant self-regulation and HRV, even though lower levels of drinking are more prevalent. Our aim was, therefore, to evaluate the impact of low levels of PAE and of maternal stress on autonomic regulation within the first days of life in infants born to a well-characterized prospectively-recruited cohort.

MATERIAL AND METHODS

Ethanol, neurodevelopment, infant and child health (ENRICH-2) study

Mothers and infants from the ENRICH-2 longitudinal prospective cohort participated in this study, following approval by the University of New Mexico (UNM) Health Sciences Center Institutional Review Board (IRB). Enrollment was conducted between 2018 and 2022. The prospective design spans from the second trimester of pregnancy until 6–9 months after birth and includes four study visits (V1–V4). The study design incorporated maternal structured interviews and biological sample collection from two prenatal visits, collection of maternal blood and urine samples, placenta, and umbilical cord blood samples at delivery (V3), a comprehensive developmental assessment at birth/first month of life, and a 6-month follow-up assessment of child development (V4).

The first two study visits (enrollment [V1] and third trimester [V2]) were conducted in the antenatal period. At V1, socio-demographic information was collected, including participant age, marital status, education, employment, and ethnic group. Health information included gravidity, parity, complications during the pregnancy, chronic health conditions, and medications. A short screener using the AUDIT-C questionnaire was used to inquire about binge drinking episodes (≥4 drinks/occasion) in the periconceptional period (2 weeks before and 2 weeks after the last menstrual period).^51–56

A Timeline Follow-Back (TLFB) interview^57,58 was used to collect information about alcohol consumption, date of last drinking, maximum number of drinks in a day, and binge drinking episodes at five different time points: the periconceptional period, and 30 days prior to each of the scheduled visits (V1-V4). One standard drink unit (SDU) was the equivalent of one 12-ounce can or bottle of beer, one 5-ounce glass of regular wine, 1.5 ounces of hard liquor, or one mixed drink with 1.5 ounces of hard liquor. The quantity and frequency reported in the TLFB were used to calculate the absolute ounces of alcohol/day.

Biomarkers were obtained at the V1 session. The ethanol biomarkers collected included γ-glutamyltranspeptidase (GGT), carbohydrate-deficient transferrin (%dCDT), phosphatidylethanol (PEth) and urine ethyl glucuronide and ethyl sulfate (uEtG/uEtS).^59,60 In addition, the pregnant participant’s blood and urine samples were collected to ascertain alcohol and illicit drug use (i.e., cocaine, methamphetamines, heroin, or ecstasy) or medication assisted therapy (e.g., methadone or buprenorphine).

Participants were assigned to the PAE or Control group using a 3-tiered screening process (see Fig. 1). Tier I consisted of the AUDIT-C questionnaire and information about binge drinking episodes in the periconceptional period. Those participants with an AUDIT-C score of ≥2 and reports of ≥2 binge drinking episodes or >13 drinks around the time of the last menstrual period were provisionally enrolled in the PAE group; those with no binge drinking and an AUDIT-C score of <2 were provisionally enrolled in the Control group. Tier II focused on alcohol use during pregnancy and required more than minimal-risk alcohol use (>13 SDU per month) based on the prospective 30-day TLFB interviews. Participants who reported greater than 13 SDU in the four TLFB or at least one binge episode during pregnancy remained in the PAE group, with the Control group consisting of individuals with no alcohol use beyond the periconceptional period (no binge episodes and no SDUs). Tier III assessed for ethanol biomarkers; Control participants had all negative biomarkers. To remain in the PAE group, the participant was required to have ≥1 binge episode or >13 drinks or ≥1 positive biomarker. Alcohol exposure levels required in the PAE group follow greater than “minimal risk” exposure levels recommended for diagnosis of PAE-related disorders.⁶¹

Fig. 1. Flowchart of eligibility for study participants.

Of the 765 participants screened, 223 participants met inclusion and eligibility criteria to proceed with the study.

Co-exposure to nicotine and marijuana was not considered a reason for exclusion in both PAE and Control groups and were assessed via self-report and urine drug screen both at enrollment (V1) and delivery (V3). The National Survey on Drug Use and Health questionnaire was used to screen for self-reported drug use. A 7-panel drug test was used to screen for basic opiates (including codeine, morphine, and heroin), expanded opiates (including oxycodone and hydrocodone), amphetamines, cocaine, phencyclidine (PCP), marijuana, and ecstasy. Those with prenatal use of cocaine, amphetamines, opioids/medication assisted therapy (methadone, buprenorphine), or ecstasy were excluded based on either self-report, positive drug test, or medical record review.

During the V2 visit, the mothers were administered the following mental health interviews or questionnaires: the Perceived Stress Scale (PSS),⁶² Generalized Anxiety Disorders-7 questionnaire (GAD-7),⁶³ Edinburgh Depression Scale (EDS),^64,65 modified Medical Outcomes Study Social Support (MOSS),⁶⁶ Posttraumatic Stress Disorder Checklist for Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (PCL-5), and Adverse Childhood Experiences (ACEs) questionnaire. Any participant with a serious mental health disorder (e.g., psychosis or schizophrenia) was excluded from the study.

V3 occurred with the infant delivery, at which time the Barratt Simplified Measure of Social Status⁶⁷ and information on any new major life events since V2 were collected from the mother. Maternal blood and urine samples were again collected at V3, and dried blood spots (DBS) were collected from the newborns via heel-lancing at the time of a clinically indicated blood collection to measure the PEth (PEth-DBS).^59,68,69 Children with severe fetal/infant anomalies were excluded from the study (e.g., major structural abnormality or complications requiring surgery or general anesthesia, such as congenital diaphragmatic hernia or gastroschisis) as well as those born <35 weeks gestational age since these problems may have resulted in similar dysregulatory outcomes unrelated to the focus of the study. Infants with a prolonged hospitalization were excluded retrospectively. Maternal or newborn administration of corticosteroids or children not living with a biological parent were also exclusionary criteria.

A total of 135 participants met all eligibility criteria after V2 and advanced through the study pipeline. Among those, 3 participants voluntarily withdrew and 1 was lost to follow-up due to delivery at another hospital, resulting in 131 subjects who completed V3 at least partially (97%). Among those, 80 (61.1%) had complete HRV data available. During the COVID-19 pandemic, UNM Hospital had restrictions on all in-person clinical research activities beginning March 17, 2020, with all restrictions being lifted at the end of May 2022, which affected in-hospital collection of HRV data by the research team. The study team was able to collect at least partial information from V3 for the majority of participants by switching to phone interviews to retain them in the cohort.

Heart rate variability

During V3, the infant EKG was administered to obtain HRV measures during the newborn heel-lance performed at 24 h post-delivery. Specifically, a BIOPAC Systems, Inc., MP150 was used to obtain the data. Three cardiac electrodes were attached to the infant’s chest, with a pneumogram transducer taped midline at the diaphragm level. The tracing was then recorded for 3 min to obtain a baseline measurement prior to heel lancing (baseline). The acute stressor was comprised of a heel lance used to obtain blood for the routine newborn screening labs and collection of a DBS card for PEth analysis. The HRV was recorded during this period (heel lance), with at least 3 min of recording obtained during the blood collection procedure. Finally, a recording was obtained for 3 min after the completion of blood collection (recovery).

The time domain measures of HRV quantify the time intervals between QRS complexes, which is the R-R interval (from R-R peaks) or N-N interval (normal-to-normal intervals between adjacent QRS complexes resulting from sinus node depolarizations). The time domain measures are calculated from the N-N intervals and include the root mean square of the successive differences (RMSSD).⁷⁰ The frequency domain measures are used to assess the spectral component of HRV and include low frequency (LF) power (0.04–0.15 Hz), high frequency (HF) power (0.15–0.40 Hz) and LF/HF ratio. The respiratory sinus arrhythmia (RSA), which measures the variation in heart rate that occur with respiration, was also obtained. Upon completion of the recordings, the electrodes and pneumogram transducer were removed using a natural oil to minimize skin irritation. The HRV measures were extracted using the QRSTool/CMETX software (Tucson, Arizona, United States) for data cleaning and Kubios software (v3.4.2; Kuopio, Finland) for analysis of the IBI (inter-beat-interval, consistent with time interval between successive ECG R-waves) data.

Statistical analysis

Demographic and clinical characteristics were summarized using means and standard deviations (SD) for continuous variables and counts (percentages) for categorical variables. Distributions for HRV measures were reviewed for normality assumptions, and HRV measures utilized in regression analyses, with the exception of heart rate, were natural log-transformed. Pearson correlation coefficients between PSS, GAD-7, EDS, and PCL-5 scores were examined.

Generalized least squares mixed effects models using maximum likelihood were created for each of the HRV measures. These models accommodate missing values in the response variable by making use of all the available information (the repeated measures). Models accounted for the repeated HRV measures for each participant, and covariance structures were selected based on the lowest AIC score. The lowest AIC scores were found for a compound symmetry covariance structure on every HRV measure except Ln LF/HF, for which an unstructured covariance structure was used. Once covariance structures were selected, restricted error maximum likelihood estimates were used. Separate univariate models were constructed to examine relations between the HRV measures and the effects of group and of HRV episode, as well as marijuana use and each of the mental health scales. Initial multivariable models for each HRV measure were then constructed to examine the interaction between groups (PAE vs. Control) and HRV episode (baseline as the reference episode). Following this, separate multivariable mixed effects models were constructed with group (PAE vs. Control), HRV episode (baseline as the reference episode), marijuana use, and each of the mental health scales. Least squares estimates from the multivariable models were used to estimate Bonferroni-adjusted pairwise comparisons between each time measurement (i.e., baseline, heel lance, recovery).

SAS statistical software (version 9.4; Cary, North Carolina, United States) was used for statistical analyses. Analyses were two-tailed, and statistical significance was determined with an alpha level of 0.05; p < 0.10 representing significance at a trend level are also reported.

RESULTS

No significant differences between the groups were observed in the maternal sociodemographic characteristics, which included maternal age, marital status, ethnic group, education level, employment status, income, and medical insurance type (Table 1, p > 0.05). The average maternal age was slightly less than 30 years. Just over half the women identified as Hispanic/Latina (56%); the majority were married or cohabitating (~70%). Over 60% of women were employed and 28% had some college or vocational school education, while about 39% had a college degree. No differences were observed between groups for income, with only 28.1% of PAE and 41.7% of Controls earning less than $30,000 per year. No significant between group differences were found for the infant characteristics and birth outcomes, including gestational age at delivery, birth weight, sex, mode of delivery, and preterm status, (Table 1, p > 0.05).

Table 1.

Characteristics of the maternal and infant study participants stratified by study group (N = 80).

Variable	Control (n = 48)	PAE (n = 32)	p value
Maternal characteristics:
Age in years (Mean ± SD)	28.8 ±5.4	29.9 ± 6.0	0.420a
Marital status:			0.800b
Single/separated/divorced	13 (27.1%)	10 (31.3%)
Married/Cohabitating	35 (72.9%)	22 (68.8%)
Ethnic Group: Hispanic/Latina	28 (58.3%)	18 (56.3%)	1.000b
Education level:			0.960b
High school or less	16 (33.3%)	10 (31.3%)
Some college or vocational school	12 (25.0%)	9 (28.1%)
College degree or higher	20 (41.7%)	13 (40.6%)
Employment status: Employedc	30 (62.5%)	20 (62.5%)	1.000b
Income:			0.250b
Under $30,000	20 (41.7%)	9 (28.1%)
$30,000–49,000	9 (18.8%)	11 (34.4%)
$50,000–69,000	7 (14.6%)	2 (6.3%)
$70,000 or over	12 (25.0%)	10 (31.3%)
Medical insurance type:			0.730b
No insurance	12 (25.0%)	5 (15.6%)
Employer-based insurance	21 (43.8%)	14 (43.8%)
Medicaid	14 (29.2%)	12 (37.5%)
Other	1 (2.1%)	1 (3.1%)
Infant characteristics:
Gestational age (Mean ± SD)	38.6 ± 1.5	38.3 ± 1.4	0.400a
Birth weight (Mean ± SD)	3312.4 ± 516.3	3255.1 ± 432.7	0.610a
Sex: female	26 (54.2%)	14 (43.8%)	0.490b
Mode of delivery:			0.920b
Vaginal-cephalic	38 (79.2%)	24 (75.0%)
Caesarean section—primary	4 (8.3%)	3 (9.4%)
Caesarean section—secondary	6 (12.5%)	5 (15.6%)
Preterm delivery (<37 weeks)	5 (10.4%)	2 (6.3%)	0.700b

AA absolute alcohol (in ounces), SD standard deviation.

^aBased on pooled variances t-test.

^bBased on Fisher’s exact test.

^cEmployment was captured at Visit 3 (birth hospitalization).

As shown in Table 2, the average amount of alcohol per day across the pregnancy was found to be 0.12 ± 0.11 ounces for the PAE group with the average amount of alcohol per drinking day during pregnancy 0.51 ± 0.27 ounces. This is considered low exposure⁷¹ or less than 1 drink/day on average. The prevalence of marijuana use was higher among participants in the PAE than the Control group (p < 0.01).

Table 2.

Maternal substance use among PAE and control participants (N = 80).

Variable	Control (n = 48)	PAE (n = 32)	p-Value
Average AA/day across periconceptional period and pregnancya (Mean ± SD)	0.002 ± 0.006	0.12 ± 0.11
Average AA/drinking day across periconceptional period and pregnancya (Mean ± SD)	0.03 ± 0.085	0.51 ± 0.27
Nicotine useb	1 (2.1%)	6 (18.8%)	0.015c
Marijuana useb	4 (8.3%)	11 (34.4%)	0.007c

AA absolute alcohol (in ounces), SD standard deviation.

^aAverage AA/day across pregnancy was determined through four TLFB calendars (periconceptual period, 30 days prior to each study visit [V1–V3]).

^bSubstance use was assessed based on three questions and covered the period from last menstrual period through to delivery (V3).

^cBased on Fisher’s exact test.

For all mental health measures except the ACEs, significant differences were observed between the PAE and Control groups (all p’s < 0.05, Table 3). Women in the PAE group had higher scores for PSS, GAD-7, EDS, and PCL-5, indicating greater levels of stress, anxiety, depression, and trauma experience. PSS, GAD-7, EDS, and PCL-5 were all highly correlated (all r’s > 0.7, all p’s < 0.01), with EDS (depression) having the highest correlation values with the other measures (r = 0.79 with GAD-7, r = 0.77 with PSS, and r = 0.74 with PCL-5).

Table 3.

Distribution of maternal stress and mental health scores among PAE and control participants (N = 80).

Variable	Control (n = 48)	PAE (n = 32)	p value
	Mean ± SD	Mean ± SD
Perceived Stress Scale	11.5 ± 7.6	16.9 ± 6.1	0.001a
General Anxiety Disorder-7	4.1 ± 3.7	7.3 ± 5.3	0.002a
PCL-5	10.2 ± 13.5	19.8 ± 17.9	0.002a
Edinburgh Depression Scale	5.6 ± 5.1	7.9 ± 4.8	0.011a
Adverse Childhood Experiences	2.1 ± 2.6	2.8 ± 3.1	0.400a

SD standard deviation, PCL-5 Posttraumatic Stress Disorder Checklist for the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition.

^aBased on Mann-Whitney test.

Figure 2 summarizes average HRV values for infants at the three HRV episode points—baseline, heel lance, and recovery. Of the 80 participants, RSA values were missing for 5 participants (5.9%). All participants had sufficient data quality (despite infant movement) to assess HRV variables including measurement for heart rate, RMSSD, LF, and HF, for at least 1 episode. The baseline episode had the most complete data (>85% for all measures), while during the heel-lance episode HR signal quality was the lowest (75–79%). HRV measures for all three episodes were available for 61.2% of participants, 24.7% had measures for two episodes, and 14.9% had measures for one episode. There were no significant differences between the PAE and Control groups for missing HRV measures at each HRV episode, nor were differences observed between the PAE and Control groups for the baseline or heel-lance HRV measures (all p’s > 0.05). At the recovery episode, infants in the PAE group had higher mean heart rate and lower Ln RMSSD, Ln LF, Ln HF, and Ln RSA values (all p’s < 0.05).

Fig. 2. Changes in HRV measures between baseline, noxious stimuli (heel lance), and recovery episodes (N = 80).

Infants in the PAE group had significantly higher mean heart rate and lower Ln RMSSD, Ln LF, Ln FR and Ln RSA values at the recovery episode. PAE prenatal alcohol exposed group; Control group. *Indicates statistically significant difference (p < 0.05) between PAE and Control groups for the recovery episode.

The univariate associations between PAE, prenatal marijuana use, maternal stress, and mental health outcomes with HRV measures are presented in Tables 5 and 6 in the Appendix. In univariate analyses, there were significant negative effects of PAE on Ln RSA and Ln HF (both p’s < 0.05), as well as a trend for a negative effect that fell short of conventional levels of statistical significance for Ln RMSSD (p < 0.10). There was also a positive effect of PAE on Ln LF/HF that fell just short of statistical significance (p = 0.06). Significant negative associations were observed for each unit increase in PSS score and Ln RSA, Ln HF, and Ln LF/HF (all p’s < 0.05). The direction of effects for GAD-7 and EDS were similar to PSS, and GAD-7, EDS, and PCL-5 were significantly associated with Ln LF/HF (p < 0.05). Marijuana use was not associated with any HRV measure.

Initial multivariable regression models were first constructed that included the group (PAE vs. Control), time (HRV episode), and the group-by-time interaction. The interaction terms were nonsignificant for all HRV measures (all p’s > 0.05) and were dropped from subsequent analyses. With respect to differences between episodes, there were significant differences for heart rate in Bonferroni adjusted pairwise comparisons between baseline and heel lance episodes (p < 0.001) and between heel lance and recovery episodes (p = 0.002). For Ln RSA, significant differences between heel lance and recovery (p = 0.032) were also found.

Table 4 summarizes the multivariable models that include PSS. In these models, PAE status was no longer a significant factor for any HRV measure (p > 0.05). There was a significant association between PSS and Ln HF (p = 0.04) as well as PSS and Ln LF/HF (p = 0.002); while the effect for PSS and Ln RSA showed a marginal trend that fell short of conventional levels of statistical significance (p = 0.096). Results of multivariable analyses with respect to other mental health scales are summarized in the Appendix (Table 5). In these models, no significant relations between mental health scales and HRV measures were noted; though relations were observed for GAD-7 and EDS with Ln LF/HF ratio of borderline statistical significance (p = 0.06).

Table 4.

Predictors of HRV measures: results of multivariable mixed effects modeling (N = 80).

	HR		Ln HF		Ln LF		Ln LFHF		Ln RMSSD		Ln RSA
	Estimate (SE)	P	Estimate (SE)	P	Estimate (SE)	P	Estimate (SE)	P	Estimate (SE)	P	Estimate (SE)	P
Intercept	123.18 (3.40)	<0.0001	3.71 (0.34)	<0.0001	4.79 (0.27)	<0.0001	1.05 (0.17)	<0.0001	2.36 (0.16)	<0.0001	3.73 (0.31)	<0.0001
Patient Status (PAE)	2.88 (3.56)	0.42	−0.32 (0.36)	0.38	−0.23 (0.28)	0.42	0.06 (0.18)	0.74	−0.20 (0.17)	0.25	−0.46 (0.34)	0.18
Time (heel lance)	8.60 (2.01)	<0.0001	−0.28 (0.20)	0.10	−0.32 (0.18)	0.07	−0.02 (0.13)	0.88	−0.04 (0.08)	0.60	−0.21 (0.16)	0.19
Time (recovery)	1.46 (1.98)	0.46	0.09 (0.20)	0.63	−0.07 (0.18)	0.69	−0.15 (0.14)	0.31	0.07 (0.08)	0.37	0.17 (0.16)	0.28
Marijuana use	3.07 (4.23)	0.47	−0.18 (0.43)	0.67	−0.13 (0.33)	0.70	0.09 (0.21)	0.66	0 (0.21)	0.99	0 (0.40)	0.99
PSS	0.25 (0.23)	0.28	−0.05 (0.02)	0.04	−0.01 (0.02)	0.44	0.04 (0.01)	0.002	−0.01 (0.01)	0.21	−0.04 (0.02)	0.096
	F-Statistic	P	F-Statistic	P	F-Statistic	P	F-Statistic	P	F-Statistic	P	F-Statistic	P
Overall time	10.24	<0.0001	1.86	0.16	1.76	0.18	0.56	0.575	0.99	0.376	2.75	0.07

PAE prenatal alcohol exposure, PSS Perceived Stress Scale, SE standard error.

DISCUSSION

In this cohort of infants with mild PAE, there was a significant association between the alcohol exposure and HRV measures during the recovery phase from an acute noxious stimulus (heel lance). Specifically, mean heart rate was higher and RMSSD, LF, HF, and Ln RSA values were all lower in infants with PAE than in unexposed Controls. As RMSSD reflects the integrity of vagus nerve-mediated autonomic healthy control of the heart,⁷² and LF and HF reflect the activities of the autonomic nervous system,³⁷ these results indicate an alteration in the ability of infants with PAE to return to a normal baseline within the timeframe needed by unexposed Controls. Importantly, the autonomic nervous system acts in a coordinated manner with the HPA axis to mediate the overall response to stressors. Indeed, these two systems are highly interconnected: they are reciprocally innervated, interact through a feed-forward mechanism (activity in either system increases activity in the other), and are both under the inhibitory healthy control of the same brain areas (prefrontal cortex and limbic system structures such as the hippocampus).⁷³ Thus, it is possible that the altered activity and regulation of the autonomic nervous system demonstrated here reflects the fact that PAE is known to alter both autonomic and HPA activity and regulation, which would result in altered interactions between these systems. It is possible that a change in the balance between autonomic and HPA activity may be key to increasing risk for adverse health outcomes.⁷³ Thus, our findings have important implications for the long-term consequences of PAE on stress reactivity and regulation as well as behavior, adaptive function, and health following PAE.^73–78 Interesting future directions would include: (a) assessing any differences in the time needed for the autonomic nervous system to recover fully to baseline, to determine whether or not the altered activity following PAE is a delay is return to baseline function; (b) measuring cortisol levels pre- and post-stress to assess possible differences in HPA responsiveness to the stressor; and (c) examining associations/interactions between these systems.

The ability to diagnose FASD following PAE remains a challenge, as the alcohol exposure may not be documented, women may be reluctant to admit alcohol use during pregnancy or forget consumption of low levels,^58,79 and the developmental impact may not be apparent very early in life. The criteria for a diagnosis of full FAS can be based on short stature and the characteristic FAS facial features and neurobehavioral impairment, which can include impairment in self-regulation.³² Thus, the utilization of HRV in the newborn period may serve as a type of biomarker to provide additional important information in diagnosing FASD.

Our study also demonstrated significant effects of maternal stress and mental health on infant autonomic regulation. Stress has been shown to impact overall health, and the PSS is a widely used survey which can be used to assess perceived stress.⁶² Maternal prenatal stress impacts fetal brain development, emotional regulation, and infant mental disorders and increases the risk of infant mortality.^77,78,80 The PSS has been shown to be appropriate for both English- and Spanish-speaking individuals in the United States,⁸¹ both of whom were included in our study. A higher score indicates higher perceived stress. Within our cohort, an increase in the maternal PSS score was associated with a decrease in infant HRV measures, indicating that the level of maternal stress may impact the autonomic nervous system of the infant. Indeed, maternal stress has been shown to negatively impact the child’s ability to self-regulate.⁴⁹ Moreover, the fetal HPA axis has been shown to be highly susceptible to programming following PAE.⁸² Thus, in addition to its effects on infant autonomic regulation, maternal stress can more directly impact fetal HPA development, with its significant long-term consequences, as noted.^74–76,83

The Generalized Anxiety Disorder module 7 (GAD-7) is a survey that has been shown to be reliable in screening, diagnosing, and determining severity of anxiety.⁸⁴ We found an association between the GAD-7 results and LF/HF ratio, which provides additional evidence that maternal stress and anxiety during pregnancy can result in alterations in the infant’s ability to self-regulate at 24 h of age. We also found an association between the maternal scores on the Edinburgh Depression Scale (EDS), a brief self-rating scale which screens for depression and has been validated in women and men,^64,85 and the infant’s LF/HF.

Finally, we can view our data through the lens of the stress-diathesis model. It is known that alcohol, in addition to its teratogenic effects, can program developing neurobiological systems, altering brain development and increasing vulnerability to later life deficits in cognitive, behavioral, and adaptive function, as well as altered stress responsiveness, self-regulatory abilities, and physical and mental health problems.^{50,82,86–89} Furthermore, children with PAE are often at increased risk for exposure to adverse and/or stressful environments during postnatal life. In the context of the stress-diathesis model, we suggest that fetal programming of stress systems by PAE may alter neuroadaptive mechanisms that play a role in mediating the stress response, thus sensitizing the organism to stressors encountered later in life, and underlying, at least partly, the increased vulnerability to deficits and disorders in multiple domains. For example, children with PAE who then experience trauma in childhood are more likely to have deficits in language, attention, memory, intelligence, and have more severe behavioral problems compared to those experiencing trauma but without PAE.⁹⁰ Thus, children with PAE are at high risk of altered neurodevelopment at birth, which places than at higher risk of further deviations from normal development when faced with later insults.

Limitations of this study include relatively overall small sample size, which can result in an inability to detect smaller effect sizes or to further examine dose-response/timing of the PAE, and challenges in collecting high quality HRV data from neonates leading to missing HRV metrics across episodes. However, despite the small sample size, changes were nonetheless observed following exposure to low doses of alcohol during pregnancy. The mixed effects modeling used may have allowed for missing repeated measures data to be analyzed, thereby helping to optimize the use of the available data. While many studies focus on higher-dose exposures (such as 7.8 drinks per occasion 1–2 days per week⁴⁴), many women report drinking low levels of alcohol in pregnancy because they believe that “only ‘strong’ alcohol and alcohol in large quantities is harmful”.⁹¹ Thus, investigating the impact of low dose exposure is both important and clinically relevant. This cohort is unusual since it represents a largely Hispanic/Latina population, with the overall educational level of the mother higher (less than a college degree) than that seen in many previous studies where the mothers had less than a high school degree for education.^44,92

CONCLUSIONS

PAE, even at low doses, can negatively impact the development of the fetus, often with lifelong consequences. This study found differences in HRV in infants prenatally exposed to alcohol, which indicated significant alterations in autonomic regulation. HRV measure alterations were also strongly associated with maternal stress, with implications for programming of fetal HPA development. Alterations in overall regulation of the stress system will have long-term consequences for infant development.

The datasets generated during and/or analyzed during the current study are not publicly available due to the lack of such data sharing acknowledgment in the consent and IRB protocol. The request for data sharing can be considered on the case-by-case basis with a formal data sharing agreement between institutions.

IMPACT:

• Previous studies have focused on effects of moderate to heavy prenatal alcohol exposure (PAE) on autonomic dysregulation, but little is known about the effects of lower levels of PAE on infant self-regulation and heart rate variability (HRV).

• Prenatal stress is another risk factor for autonomic dysregulation.

• Mild PAE impacts infant self-regulation, which can be assessed using HRV.

• However, the effect of prenatal stress is stronger than that of mild PAE or other mental health variables on autonomic dysregulation.