Association between placental toxic metal exposure and NICU Network Neurobehavioral Scales (NNNS) profiles in the Rhode Island Child Health Study (RICHS)

Abstract

Background –

Prenatal exposure to heavy metals has been linked to a variety of adverse outcomes in newborn health and later life. Toxic metals such as cadmium (Cd), manganese (Mn) and lead (Pb) have been implicated to negatively affect newborn neurobehavior. Placental levels of these metals may provide additional understandings on the link between prenatal toxic metal exposures and neurobehavioral performances in newborns.

Objective –

To evaluate associations between placental concentrations of toxic metals and newborn neurobehavioral performance indicated through the NICU Network Neurobehavioral Scales (NNNS) latent profiles.

Method –

In the Rhode Island Child Health Study cohort (n=625), newborn neurobehavioral performance was assessed with NNNS, and a latent profile analysis was used to define five discrete neurobehavioral profiles based on summary scales. Using multinomial logistic regression, we determined whether increased levels of placental toxic metals Cd, Mn and Pb associated with newborns assigned to the profile demonstrating atypical neurobehavioral performances.

Results –

Every doubling in placenta Cd concentration was associated with increased odds of newborns belonging to the atypical neurobehavior profile (OR: 2.72, 95% CI [1.09, 6.79]). Detectable placental Pb also demonstrated an increased odds of newborns assignment to the atypical profile (OR: 3.71, 95% CI [0.97, 13.96]) compared to being in the typical neurobehavioral profile.

Conclusions –

Toxic metals Cd and Pb measured in placental tissue may adversely impact newborn neurobehavior. Utilizing the placenta as a prenatal toxic metal exposure biomarker is useful in elucidating the associated impacts of toxic metals on newborn health.

1. Introduction

Exposure to environmental contaminants is an important public health concern especially among children. Toxicants, such as heavy metals, are persistent in the environment from both natural and anthropogenic activities. The general population is often exposed, voluntarily and involuntarily, through ambient air, drinking water, dietary or industrial sources. Due to the ubiquity of heavy metals in the environment, exposure to these contaminants is often inevitable in the susceptible children’s population.

A growing body of evidence has documented heavy metals contributing to a wide variety of adverse outcomes including decreased fetal growth and length of gestation, low birth weight, and cognitive and behavioral deficits in children.^1–5 As exposure to even low levels of toxic metals during critical developmental windows have been linked to detrimental neuropsychological outcomes, it is imperative that we investigate the underlying association between prenatal heavy metal exposure and neurobehavioral outcomes in newborns at the earliest time point, as earlier prevention and intervention strategies can be the most successful.^6–8

Although much work on the mechanistic toxicology of toxic metals has focused, appropriately, on impacts to the developing brain, there is a growing body of literature suggesting an important role for the placenta and its functions in neurodevelopment.^9–11 Besides its role in nutrient, waste, gas, and water transport, the placenta also produces a variety of neurotransmitters and acts to metabolize maternal hormones in support of healthy fetal brain development.¹² These data also suggest that stressors that can impact fetal neurodevelopment may do so, at least in part, through impacts on the placenta.^13–15

Cadmium (Cd) is considered a toxic metal and has been widely studied for its adverse impacts on human health.¹⁶ Aside from the commonly known effects on kidney functions and attribution to various types of cancer, Cd exposure has also been linked to neurologic, developmental and cognitive impairments in the vulnerable children’s population.^1,16,17 Cd is known to accumulate in the placenta, and that its pregnancy-related toxicity is due to impacts on the placenta including eliciting oxidative stress, interfering with the transfer of essential metals to the fetus and impairing the developmental progress of the fetus during the critical gestational period.^18–21 This would also suggest that utilizing the placenta as a biomarker when examining its health effects can provide important information.

Lead (Pb) is an extensively studied developmental toxicant and children are the targeted population of concern due both to higher exposure scenarios and the susceptibility of the developing brain to the exposure.^22–24 Even at relatively low levels, exposure can lead to detrimental effects on children’s development, thus suggesting that there may be no “safe” limit for Pb exposure.^25–27 Although Pb can be transferred through the placenta to fetal circulation, detected levels in placenta may indicate high in utero Pb exposure and result in detrimental effects to the developing fetus.^28,29

Manganese (Mn) is an essential element and is commonly found in the environment.³⁰ Aside from natural sources, dietary sources such as rice and whole grains contain the highest Mn levels.^30–32 However, Mn is additionally recognized as a toxicant with^32,33 excessive Mn targeting the developing central nervous system (CNS) and contributing to neurological disorders.^32,34 During pregnancy, Mn can accumulate in the placenta and through impacts to the placenta may elicit adverse effects on the developing CNS of the fetus.

Numerous studies have investigated the association between prenatal Cd, Pb or Mn exposure and neurobehavioral outcomes in young children. However, with different approaches of neurobehavioral assessments and varying biological matrices available for measuring prenatal heavy metal exposure, the exact relationship between metal exposure and neurobehavioral-related outcome remains uncertain. The placenta can serve as a useful biomarker for measuring prenatal exposure, and to our best knowledge, there are still research gaps in associating placental metal levels and neurobehavioral deficits.¹⁷ Therefore, our study goal is to quantify Cd, Mn and Pb levels in the placenta and examine the hypothesis that placental trace metal concentrations are related to atypical neurobehavioral performance identified at birth.

2. Methods

2.1. Study Population

The infants included in this analysis are from the Rhode Island Health Study (RICHS), a hospital-based birth cohort established at the Women and Infants Hospital in Providence, Rhode Island, USA. The RICHS cohort recruited mothers of term infants (≥ 37 weeks) oversampled for infants that were SGA (small for gestational age) and LGA (large for gestational age), while recruiting AGA (adequate for gestational age) infants matched on sex, gestational age (within 3 days), and maternal age (within 2 years) to those at the extremes. Exclusion criteria of RICHS included mothers younger than 18 years of age, pregnancies resulting in preterm birth (< 37 weeks), or infants born with congenital or chromosomal abnormalities. A total of 840 mother-infant pairs were enrolled into the study. Medical records were used to collect anthropometric and clinical data, including birth weight, gestational age, depression and anxiety/panic/obsessive compulsive disorder history. Interviewer-administered questionnaires were used to obtain information on sociodemographic, behavioral, and exposure histories.

2.2. NNNS Assessment

The NICU Network Neurobehavioral Scale (NNNS) is a standardized assessment aimed to comprehensively assess newborns’ neurologic integrity, behavioral functioning and signs of stress.³⁵ The assessments results in 13 summary scores, including habituation, attention (orientation), arousal, self-regulation, handling, quality of movement, excitability, lethargy, non-optimal reflexes, asymmetry reflexes, hypertonicity, hypotonicity, and stress/abstinence.

In RICHS, NNNS was administered after the first 24 hours of life, and prior to discharge by certified psychometrists. Of the 840 enrolled infants in RICHS, 625 infants (74%) were assessed with NNNS (Fig. 1). For the habituation construct, as the newborn was required to be asleep for assessment, information was not collected for 54.9% of the infants.³⁶ As a result, we only included the remaining 12 summary scores of NNNS for the statistical analyses.

Fig. 1.

Analysis strategy: 840 mother-infant pairs recruited in the RICHS study had available demographic information. Neurobehavioral performance was assessed via NNNS after 24 hours of birth for 625 infants (74%), and heavy metal levels were analyzed in 192 placenta samples.

2.3. Metal Assessment

Placental parenchyma tissue suitable for trace metals analysis was collected from 192 participants in RICHS within 2 hours of delivery (Fig. 1). Biopsies from approximately 1–2cm from the cord insertion site, free of maternal decidua, were excised, rinsed, and stored in trace element-free tubes until further examination. Placental concentrations of twenty-four metals, including Cd, Mn, and Pb were quantified at the Dartmouth Trace Elements Analysis Core using ICP-MS protocols as described in detail elsewhere.³⁷ Briefly, after samples were transferred and brought to room temperature, HNO₃/HCl (Optima^™) solution was added based on sample wet weight. Following EPA method 3050B, samples were then digested via microwave, and 0.25–0.35 ml of H₂O₂ was added to each sample tube. Quality control measures included the analysis of standard reference material (NIST 1566b, oyster tissue), initial and continuing calibration verification, and the use of fortified blanks, analytical duplicates and spikes. The detection limits (LOD) for placenta Cd, Mn and Pb were 2.12 ng/g, 10.61 ng/g and 2.12 ng/g, respectively.

2.4. Statistical Analysis

We compared demographic characteristic differences between the RICHS study population, and the subset of participants with both placental heavy metal and NNNS data with chi-square test and t-test.

Descriptive statistics for all NNNS summary scales were reported, with higher score on each scale indicating a higher level of the measured construct. Depending on the measured aspect of neurobehavior, higher scores may either indicate better or worrisome performance. For instance, newborns with higher scores in the attention scale were more alert and able to follow visual stimulations with eyes and head during assessment. Higher level of the quality of movement scale indicated predominantly smooth movements and little jerkiness or startles. On the contrary, newborns with high scores in the lethargy or stress/abstinence scales exhibited more signs of the measured construct, which are less optimal.

We used latent profile analysis (LPA) to understand the underlying NNNS summary score patterns in the RICHS study population. According to Liu et al., LPA assumes the study population consists of several subgroups that can be labeled as latent profiles or classes.³⁸ The purpose of using LPA for the NNNS scores was to produce discrete profiles with minimized heterogeneity within a profile and maximized heterogeneity across different profiles. In order to obtain an optimal number of profiles based on the NNNS summary scores in the study population, we fitted multiple models with different number of profiles, and the model with the most appropriate fit statistics was chosen as the number of profiles for RICHS. Referring to information-criteria based fit statistics, the model with the lowest log-likelihood and Bayesian Information Criteria (BIC) values was preferred. Entropy value greater than 0.8 is ideal as this would indicate greater classification accuracy of the subjects. Additionally, profile size will also be taken into consideration, as the smallest profile should still include at least 5% of the study population.³⁹

Heavy metal levels and NNNS summary scores were compared across profiles with ANOVA analyses. The association between placental heavy metal levels and newborn neurobehavioral performances indicated through profiles was assessed using multinomial logistic regression. Samples with Cd levels below the detection limit were assigned a value of LOD divided by the square root of 2. We applied log2-transformed Cd and Mn levels in the models for approximation to normal distribution, and for interpretation purposes. Alternatively, 51.6% samples were flagged as <LOD for Pb levels, so we then treated placental Pb as a binary variable (non-detect vs. detectable) in the regression models. Regression models were further adjusted for covariates collected from medical records and in-person questionnaires administered by interviewers and were a priori determined in reference to previous RICHS studies. Adjusted covariates included infant sex, maternal age, maternal race (white or not white), maternal body mass index, and educational attainment (dichotomized into obtained high school or less versus more than high school education). We also ran sensitivity analysis to test whether model results were robust when tobacco smoking status during pregnancy was included.

Statistical analyses were conducted using R version 3.5.1, and LPA analysis was performed with Mplus version 8.4.

3. Results

3.1. Study population

Demographic and gestational characteristics of the RICHS cohort are shown in Table 1. Of the 625 RICHS newborns with NNNS information, 51.4% were females and 57% were born under the adequate for gestational age (AGA) birth weight category. Maternal participants were predominantly white (70.6%) and had obtained at least some post high school education (73.4%). The subset of infants with both NNNS and placental heavy metal information (n=192) showed similar distribution for most of the demographic characteristics. 47.4% of the subset of infants were females, 73.4% mothers were white, and 75.5% had some post high school education. Average gestational age in the RICHS study population and subset were both around 39.3 weeks. Birth weight category distribution was significantly different between the full cohort and sub-cohort (p=0.001), and birth weight in the subset (mean=3644 g) was significantly higher than the entire study population (mean=3477 g; p=0.003).

Table 1.

Demographic and gestational characteristics.

	Newborns with NNNS data (N=625)	Newborns with NNNS and placental metal data (N=192)
	n (%)
Infant Gender
Female	321 (51.4%)	91 (47.4%)
Male	304 (48.6%)	101 (52.6%)
Birth Weight Categorya
SGA	123 (19.7%)	29 (15.1%)
AGA	356 (57.0%)	93 (48.4%)
LGA	146 (23.4%)	70 (36.5%)
Delivery Method
Cesarean section	317 (50.7%)	110 (57.3%)
Vaginal	308 (49.3%)	82 (42.7%)
Maternal Race
White	441 (70.6%)	141 (73.4%)
Other	167 (26.7%)	47 (24.5%)
Unknown	17 (2.7%)	4 (2.1%)
Infant Race
White	408 (65.3%)	123 (64.1%)
Other	201 (32.2%)	64 (33.3%)
Unknown	16 (2.6%)	5 (2.6%)
Maternal Education Status
No more than high school	166 (26.6%)	47 (24.5%)
Some post-high school	459 (73.4%)	145 (75.5%)
Maternal Smoking Status
Yes	64 (10.2%)	26 (13.5%)
No	555 (88.8%)	165 (85.9%)
Unknown	6 (1.0%)	1 (0.5%)
	Mean ± SD
Birth weight (grams)	3477 ± 664.75	3644 ± 680.37
Gestational age (weeks)	39.34 ± 0.96	39.31 ± 0.95
Maternal age (years)	29.48 ± 5.49	29.79 ± 5.63
Maternal BMI (kg/m2)	26.71 ± 6.96	27.48 ± 7.14

^aSGA: small for gestational weight; AGA: adequate for gestational weight; LGA: large for gestational weight

3.2. NNNS profiles

Descriptive statistics of the summary scales of NNNS are shown in Supplement Table 1 and were used to identify potential outliers. Extreme values were found for stress abstinence, non-optimal reflexes, and asymmetric reflexes scales, but no individuals in this study population were consistently outliers for all of the NNNS summary scales.

To determine the most appropriate number of profiles in this study population, we fit LPA models with 2 to 6 profiles. As the fitted profile numbers increased from 2 to 6 profiles, BIC values decreased (Supplement Table 2). Entropy values for all the models were all greater than 0.8, which was preferred, and showed better accuracy in classifying participants into different profiles. Additionally, profile size was used to determine the optimal number of profiles as it is recommended that the smallest profile should not be smaller than 5% of the study population. Despite having the lowest BIC value, one of the profiles generated from the 6-profile model consisted of only 29 participants (4.6%). Based on fit statistics, profile size, and average class probabilities, the 5-profile model showed the best fit for the RICHS cohort.

Fig. 2 demonstrated distinct NNNS patterns of the five profiles based on the standardized summary scores as the original scores were not on the same scale. Mean and standard deviation of the 12 summary scores by profiles are shown in Supplement Table 3. The largest profile (Profile 2) consisted of 172 (27.5%) participants, while the smallest profile (Profile 5) included 45 (7.3%) subjects. Compared to other profiles, Profile 5 infants showed the most extreme scores, indicating atypical neurobehavior compared to infants in other profiles. Infants in this profile were characterized with the highest arousal, excitability, hypertonicity and stress abstinence signs, and the lowest quality of movement, regulation and non-optimal reflexes. Profile 4 (N=124, 19.8%) infants showed more signs of lethargy, hypotonicity, non-optimal reflexes and asymmetric reflexes, and lowest attention and arousal. Compared to Profiles 4 and 5, Profile 3 infants (N=157, 25.1%) required more handling, but behaved on average for most of the summary scores. The largest group, Profile 2, showed relatively average performances for most of the summary scores with the exception of the lethargy scale, which was the lowest across all profiles. Infants in Profile 1 (N=127, 20.3%) showed the highest attention, quality of movement, and regulation, and lowest stress abstinence signs, along with less handling, excitability, and hypertonicity. Given the summary scale pattern indicating most scales around the mean, in the following regression models, Profile 2 served as the reference profile.

Fig. 2.

Five NNNS profiles (N=625). NNNS summary scale z-scores across all five latent profiles as indicated via LPA among all newborns in the RICHS study population. Profile 5 (black) demonstrates atypical neurobehavior and Profile 2 (green) represents typical neurobehavioral performance.

3.3. Placental heavy metal concentrations

The mean, standard deviation, minimum and maximum values of placental Cd, Mn and Pb are shown in Supplement Table 4. We further examined log₂-transformed placental metals concentration and detectable placental Pb distribution by the five NNNS profiles (Fig. 3). Placental Cd and Mn did demonstrate an elevation in Profile 5, although the differences across profiles were not considered statistically significant. We also observed a greater proportion of detectable Pb amongst placenta from individuals classified in Profile 5.

Fig. 3.

Placental heavy metal concentration and distribution across five NNNS profiles. For placental Cd (A) and Mn (B), y-axis shows log₂-transformed concentrations, while the x-axis demonstrates NNNS Profiles 1–5. For placental Pb (C), stacked bar plot represents the percentages of detectable and non-detectable distributions across the five NNNS profiles (x-axis) in the RICHS study population.

3.4. Association between heavy metal and profiles

In line with the bi-variate results, in the unadjusted multinomial logistic regression results (Table 2), detectable placental Pb also demonstrated an increased odds of newborns belonging to the atypical Profile 5 (OR: 3.12, 95% CI [0.89, 10.97]) with Profile 2 as the comparison group. In adjusted regression models, compared to belonging in Profile 2 with typical neurobehavior, there were increased odds of newborns being placed in the atypical neurobehavior Profile 5 with every doubling increase in placenta Cd level (OR: 2.72, 95% CI [1.09, 6.79]). Additionally, newborns with detectable placental Pb levels also demonstrated an increased odds of belonging to Profile 5 (OR: 3.71, 95% CI [0.97, 13.96]), although in all models the 95% CI are wide suggesting potentially unstable results. Alternatively, increased placenta Mn level was not associated with increased odds of newborns belonging to the atypical Profile 5 (OR: 1.16, 95% CI [0.28, 4.92]). As a sensitivity analysis, we fit models with maternal smoking status as a covariate (Table 2: Adjusted^b), noting that the number of smokers in this population is small and the number of smokers within any profile was limited. These models indicated some attenuation of the effect size for the associations between placental Cd and Pb with Profile 5. In models examining membership in Profile 5 versus all other profiles (Supplement Table 5), we observed an association between detectable placental Pb levels and increased odds of atypical neurobehavioral performances in newborns (OR: 3.94, 95% CI [1.15, 13.46]), as well as increased odds for Profile 5 membership associated with increasing Cd concentrations (OR: 2.39, 95% CI [1.03, 5.52]).

Table 2.

Odds ratio (95% CI) from multinomial regression models (reference group: Profile 2).

	Profile 1	Profile 3	Profile 4	Profile 5
Unadjusted
log2 Cd	1.30 (0.74, 2.31)	0.95 (0.55, 1.64)	0.90 (0.51, 1.57)	1.70 (0.78, 3.66)
log2 Mn	0.64 (0.24, 1.70)	0.97 (0.39, 2.45)	0.87 (0.34, 2.25)	2.38 (0.65, 8.63)
detectable Pb	0.73 (0.31, 1.68)	0.91 (0.41, 2.03)	0.90 (0.39, 2.05)	3.12 (0.89, 10.97) *
Adjusteda
log2 Cd	1.55 (0.82, 2.92)	1.02 (0.57, 1.84)	1.08 (0.59, 1.98)	2.72 (1.09, 6.79) **
log2 Mn	0.57 (0.19, 1.66)	0.79 (0.29, 2.15)	0.92 (0.33, 2.58)	1.16 (0.28, 4.92)
detectable Pb	0.93 (0.38, 2.29)	0.96 (0.41, 2.21)	0.88 (0.37, 2.09)	3.71 (0.97, 13.96) **
Adjustedb
log2 Cd	1.57 (0.83, 2.99)	1.03 (0.57, 1.86)	1.11 (0.60, 2.03)	2.27 (0.90, 5.70) *
log2 Mn	0.57 (0.20, 1.67)	0.80 (0.30, 2.16)	0.93 (0.33, 2.63)	1.05 (0.24, 4.61)
detectable Pb	0.95 (0.38, 2.35)	0.97 (0.42, 2.25)	0.91 (0.38, 2.20)	3.42 (0.88, 13.32) *

^aAdjusted for infant gender, maternal age, maternal race, maternal BMI, education status

^bAdjusted for infant gender, maternal age, maternal race, maternal BMI, education status, smoking status during pregnancy

^*p<0.1;

^**p<0.05

4. Discussion

Our study explored placental concentrations of putative neurotoxic trace metals and neurobehavior assessed by NNNS profiles. By categorizing our study population based on the NNNS using the LPA method, and after controlling for multiple covariates, we observed an association between increased placental Cd levels and higher odds of newborns belonging to the atypical neurobehavior profile (Profile 5) compared to them being placed in the typical Profile 2. Detectable placental Pb was also associated with increased odds of newborns being placed in the atypical neurobehavior profile compared to those with placental Pb levels below the detection limit.

Metal concentrations measured in this present study were generally consistent with levels detected in other study populations. Cd levels measured in the RICHS placenta samples were towards the lower end of the reported range of levels worldwide.⁴⁰ A review reported that the average concentration of placental Cd was around 4 ng/g in non-exposed environments.^40,41 Our reported arithmetic placental Cd mean at 4.56 ng/g (±SD 2.58) was lower compared to levels measured in some countries outside the US, but slightly greater than levels measured in a study in North Carolina.^42–44 The range of placental Mn level in RICHS was much wider than the detected range in a birth cohort study in Spain.⁴⁵ Amongst those with detectable Pb levels, placental concentrations averaged at 4.49 ng/g (±SD 3.89), which was lower than the reported value from several study populations across the world, but comparable to the value reported in the New Hampshire Birth Cohort study.^37,40,42,46

Cd is a known developmental toxicant as established by numerous studies, and common sources of Cd exposure are diet, smoking during pregnancy or industrial contamination in residential environment. The RICHS population had relatively low prevalence of women who smoked during pregnancy (10.2%), and the study setting was not occupational in nature, thus the main source of Cd exposure measured in the placenta was most likely due to diet.⁴⁷ However, further understanding of dietary Cd sources in RICHS is limited as we did not obtain dietary information from the participating women.

Proper placental functions are crucial for normal fetal growth and development. During pregnancy, the placenta acts as a barrier and is thought to protect the developing fetus from Cd exposure by limiting transfer to fetal circulation. Nonetheless, Cd levels were detectable in cord blood and newborn serum in previously published studies.^48,49 In our study, we investigated prenatal Cd exposure using the placenta as a biomarker and observed a significant association between Cd and the atypical Profile 5, characterized by poorer neurobehavioral performances, such as increased signs of stress and excitability. The MOCEH study in Korea found no association between Cd measured in maternal blood and neurodevelopment measured at 6 months.⁵⁰ The HOME study also did found no significant association between maternal urinary Cd and cognitive and behavioral outcomes in 1–8-year-old children.⁵¹ Additionally, no relationship between blood Cd and neurodevelopment was observed in children 2 years of age in the TLC study.⁵² Limited numbers of epidemiologic studies have investigated the potential link between prenatal Cd exposure and fetal development through impacts on the placenta. However, in line with our findings, other studies have identified inverse associations between prenatal Cd exposure measured through maternal blood, cord blood or urine, and verbal and performance IQ in children.^1,53,54 As we further explore prenatal Cd exposure in the placenta, the present finding adds to the understanding of the potential for adverse impacts from prenatal Cd exposure on fetal development.

We further observed an association between detectable Pb and assignment to the atypical profile for newborns in the RICHS population. Pb is well-documented for its developmental neurotoxicity in children.^6,24,55 Our results are in agreement with a series of studies that demonstrated the negative effects on neurodevelopment from early-life Pb exposure. Postnatal dentine Pb was linked to increased behavioral problems in Mexican children between the ages of 8–11.⁵⁶ In addition, prenatal and early postnatal exposure to Pb have also been found to be associated with decrements in IQ and compromised neuropsychological function in children.^{8,23,25,57,58} Contrary to these findings, Freire et al. measured placental Pb in the INMA cohort as we did in the present study and found no association with neurodevelopmental outcomes at 4–5 years of age.⁴⁵ Additionally, Taylor et al. did not observe association between placenta Pb levels and motor skills in 7-year-old children.²⁷ Among infants, the MOCEH study also found no association between Pb exposure measured early in pregnancy and mental (MDI) and psychomotor (PDI) development index scores, but Pb levels during late pregnancy period was found to be linked to lower MDI scores assessed at 6-month of age.⁵⁰ Thus, whether early indicators of neurodevelopmental performance among newborns associate with placental Pb will need to be confirm or refuted by further studies.

We also investigated Mn exposure and newborn neurobehavior in this study population but did not find a significant association between increased placental Mn levels and higher odds of newborns belonging to the atypical Profile 5. Unlike Cd and Pb that are classified as toxic metals, Mn is recognized as an essential nutrient crucial for development and growth, though excess levels have been linked to cognitive and behavioral issues early in life, at 6 months in the MOCEH study and in children at 2 years of age from a study in Taiwan.^30,32,59,60 A meta-analysis found a 50% increase in Mn levels was associated with decreased IQ points in children aged 6–8 years.¹⁷ The CHAMACOS cohort also linked increased Mn exposure to poorer behavioral performance⁶¹ In contrast, the same study found improved memory and cognitive function in older boys to be associated with higher Mn levels. In agreement with our study, a French study found no association between Mn exposure measured in the placenta and cognitive scores at age 3 or age 6.⁶² The mean Mn level in our study cohort was similar to the French study (0.095 ug/g vs. 0.10 ug/g), thus it is possible that placental Mn at this level was not high enough to negatively impact neurobehavior.

While our findings on placental Cd and Pb and atypical neurobehavioral phenotype generally agreed with prior studies, discrepancies may have arisen from the type of biologic matrices used. Maternal urine, maternal blood, cord blood, hair and toenails have all been used to determine the link between Cd, Pb or Mn levels and children’s neurobehavior, with fewer studies of placenta. Thus, the metals’ characteristics, such as Cd’s accumulative nature in the placenta and Pb’s ability to pass through the placenta along with the time point in which biological matrices were collected, or the study population’s exposure environment and cultural patterns, could have resulted in differences in the direction and strength of association between neurodevelopment and metal exposures across studies.

We were able to distinguish NNNS score pattern differences across five profiles and identified typical and atypical profiles based on neurobehavior characteristics. Initially designed to study effects of prenatal drug exposure on child outcomes, the NNNS assessment utilizes standardized scales to examine newborns’ neurobehavioral performances.^36,63 From analyzing NNNS results from a large random sample of clinically healthy newborns, Fink et al. showed that NNNS scales ranging between the 10^th to the 90^th percentile would indicate normative neurobehavior.⁶⁴ Researchers have since successfully applied NNNS assessments to understand neurobehavior potentially associated with different environmental exposures in healthy full-term infants.

Like others, we used the LPA approach to analyze NNNS data. We observed similarities in the identified profiles to those previously of Liu et al. in the Maternal Lifestyle Study (MLS), a multicenter longitudinal study designed to understand effects of illicit drugs during pregnancy on the mother, fetus and infant.^38,65 The similarities of the NNNS patterns across the profiles of these two populations provides evidence that NNNS assessment and profiling method can be used to characterize neurobehavioral performances in low-risk, healthy infants and to investigate the determinants of these profiles.

NNNS scores evaluated within 24–72 hours of birth have prospectively predicted infant temperament and neuropsychological characteristics in toddlers and preschool-aged children.^38,66,67 Thus, the application of the NNNS measurements may allow early insight in the newborns’ neurobehavioral performances which could inform ongoing monitoring or early interventions to those categorized into the atypical profiles. It is possible that by ascertaining the newborn’s neurobehavioral performance before hospital discharge, rather than waiting for adverse neurobehavior to manifest, we would be able to provide interventions within the early developmental period.

Advantages of this study include understanding prenatal metal exposure using the placenta as a biomarker. As prenatal metal exposures are known to elicit adverse impacts on the developing fetus, measuring placental levels of Cd, Mn and Pb may offer new understandings on prenatal exposure characteristics and associated neurobehavioral performance within days of life. With the LPA method, we were able to generate discrete neurobehavioral profiles, and the relatively large cohort size (N=625) in RICHS allowed adequate profile sizes with distinct NNNS patterns comparable to prior work in another study population, suggesting our findings are generalizable beyond the study region. We were able to identify atypical neurobehavior if newborns were assigned to Profile 5, though the profile size is considerably smaller than the other profiles. This is likely to due to RICHS being comprised of primarily low-risk and healthy participants, so there are relatively fewer extreme cases of neurobehavioral issues among the newborns.

For this present study, we analyzed individual metal’s contribution to atypical NNNS score patterns, but it is likely that the population was concurrently exposed to multiple metals. To better address how these developmental toxicants impact child’s neurobehavior, additional research considering metal exposures as a mixture and identifying the major contributor(s) driving the impacts of metal mixture on neurobehavior is needed. Another limitation of this study is the lack of information on exposure source of the participants. Other than smoking, a common source of exposure to Cd in the general population is dietary intake, such as shellfish or other Cd-contaminated foods.¹⁶ The general population is also primarily exposed to Mn through food and water. It is likely that with the lack of dietary data, we are restricted to fully address the association between metal exposure and neurobehavior, as certain dietary components may lead to increased metal exposure, but are also considered beneficial towards development. Aside from dietary intake, inhalation of particulate matter containing Mn or dermal contact with Mn-contaminated air, water, and soil are all possible sources of Mn exposure.³⁰ While the overall metal exposure levels in RICHS are lower compared to other studies, future work examining metal levels based on the different exposure sources could further clarify the adverse impacts of metals on neurobehavior.

5. Conclusion

In this study, we found that placental toxic metals including Cd and Pb negatively affected neurobehavioral performance in newborns in a generally healthy population as indicated through the NNNS. Our findings also highlight the importance of the placenta in newborn health and the utility of measuring of placental metal concentrations to evaluate child health outcomes.