Abstract
Selenium (Se) is an essential trace element involved in various biological processes, including neurodevelopment. Available literature indicates that both Se deficiency and excess may be detrimental to health. It is also known that Se can cross the placenta from maternal to fetal circulation. To date, the role of maternal Se status in child long-term neurodevelopment is largely unexplored. This study investigated the temporal and dose-response associations between maternal Se status and child risk of neurodevelopmental disorders including autism spectrum disorder (ASD) and attention deficit hyperactivity disorder (ADHD). It consisted of 1550 mother-infant dyads from the Boston Birth Cohort. Maternal red blood cell (RBC) Se levels were measured in samples collected within 72 hours of delivery (biomarker of 3rd trimester Se status). Pediatric neurodevelopmental diagnoses were obtained from electronic medical records. Data analyses showed that maternal RBC Se levels were positively associated with child risk of developing ASD, with an adjusted odds ratio of 1.49 for ASD (95% CI: 1.09, 2.02) per IQR increase in Se. There was also a positive association between maternal Se and ADHD (OR 1.29, 95% CI: 1.04, 1.56, per IQR increase in Se). These associations remained robust even after adjusting for pertinent covariables; and there was no significant interaction between Se and these covariables. Our findings suggest that prenatal exposure to high maternal Se levels may adversely affect child neurodevelopment. Our findings warrant further investigation; if confirmed, optimizing maternal prenatal Se levels may be necessary to maximize its health benefits while preventing undue risk.
Lay Abstract
Selenium (Se) is an essential nutrient for the health of the pregnant mother and her baby. While Se can readily cross the placenta from maternal to fetal circulation, little is known about maternal Se status on her child’s neurodevelopmental outcomes. We studied over 1500 mother-child dyads from birth to school age of the child. We found that babies born from mothers with high blood Se levels may be at increased risk of developing autism spectrum disorder or attention deficit hyperactivity disorder. Given this is the first study of the kind, more study is needed to confirm our findings.
Introduction
Autism Spectrum Disorder (ASD) is a complex neurodevelopmental disorder that affects approximately 1 out of 54 children in the US (Center for Disease Control and Prevention [CDC], 2016). Despite its increasing incidence, we currently lack curative therapies for ASD, making ASD a particularly challenging disorder in Pediatrics (Bjørklund et al., 2020). While the exact pathophysiology of ASD is unknown, there is growing evidence that increased oxidative stress and neuroinflammation in the brain may be contributory to ASD pathogenesis (Pangrazzi, Balasco, and Bossi, 2020). In particular, existing studies have shown that children with ASD have an imbalance in glutathione levels and glutathione reserve capacity (Bjørklund et al., 2020).
Selenium is an essential trace element that acts as a cofactor for glutathione peroxidase (also known as selenoprotein), an antioxidant enzyme, which protects the body from free radicals and carcinogenic factors (Kielczykowska, Kocot, Pazdzior & Musik, 2018). However, selenium is known for its narrow safety margin: both its deficiency and excess have been linked to adverse outcomes (Kielczykowska, Kocot, Pazdzior & Musik, 2018). For instance, selenium deficiency has been associated with congestive cardiomyopathy (Keshan disease), deformity of bones and joints (Kashin-Beck disease), as well as increased susceptibility to viral infections. Selenium may have a neuroprotective impact in those with Alzheimer’s disease (Varikasuvu, Prasad V, Kothapalli & Manne, 2019). On the other hand, excess selenium has been associated with neurologic changes, increased risk of Type 2 Diabetes, and damage to the gastrointestinal tract (Kieliszek, 2019). While acute selenium toxicity is rare, there have been cases of individuals who presented with symptoms of acute toxicity after ingesting inappropriately dosed selenium supplements (MacFarquhar et al., 2010).
This study sought to address an important knowledge gap. Although it is known that selenium can cross placenta from maternal to fetal circulation, the role of maternal selenium status in child long-term neurodevelopment is largely unexplored. Thus, this study serves as a hypothesis testing and hypothesis generating. We propose that despite being an essential nutrient for neurodevelopment, selenium may be neurotoxic at high concentrations. To better understand the role of selenium on neurodevelopment, we conducted a prospective birth cohort study to investigate whether maternal red blood cell (RBC) selenium levels are associated with childhood risk of ASD, along with other related neurodevelopmental outcomes, specifically, attention-deficit hyperactive disorder (ADHD) excluding ASD, and other developmental disorders (other DD) excluding ASD and ADHD, all of which are compared to neurotypical children (NT, as reference group). We also examined whether maternal factors including maternal stress, overweight/obese or Diabetic (OWO/DM), maternal cholesterol levels, and nutrient status (i.e., plasma folate and vitamin B12 levels) could affect the selenium-outcome associations.
Methods
Study population
We used the data from the Boston Birth Cohort (BBC), which consists of 3,165 mother-infant dyads who were enrolled at birth and remained in the follow-up study from 1998 to 2018. The BBC is a predominantly urban, low-income minority population in the US. Details of the BBC recruitment and postnatal follow-up have been published previously (Wang et al., 2014; Wang et al., 2002; Li et al., 2016).
Briefly, mothers who delivered a single live birth at Boston Medical Center (BMC) were eligible for participating in this study. Pregnancies resulting from in vitro fertilization, multiple-gestation pregnancies, deliveries induced by maternal trauma, or newborns with substantial congenital anomalies were not eligible for enrollment. As shown in the study flowchart, of the 3,165 mother-infant dyads enrolled in the BBC postnatal follow-up study, 1,550 had sufficient maternal red blood cell (RBC) samples for selenium assay (Supplemental Figure 1). The Institutional Review Board (IRB) of the Boston University Medical Center and Johns Hopkins Bloomberg School of Public Health approved the BBC study. Informed consent was obtained from each participant under the IRB approved protocol.
Selenium in maternal RBC and other biomarkers in maternal plasma
Although selenium can also be measured in other type of biospecimens, including hair (De Palma, Catalani, Franco, Brighenti & Apostoli, 2012) and nails (Lakshmi Priya and Geetha, 2010), RBC is the primary storage sites for selenium. More importantly, in our prior study (Chen et al., 2014), we have demonstrated that maternal selenium readily crosses the placenta and that the correlation between maternal RBC selenium and cord RBC selenium is greater than that between maternal plasma selenium and cord plasma selenium. Another advantage of RBC selenium is that it reflects a long-term selenium status compared to plasma (RBC lifespan is about 120 days) and is free from hemodilution during pregnancy. In addition, a focus of maternal selenium allows early risk assessment and prevention during critical fetal development period.
In this study, maternal selenium levels in RBCs collected within three days after delivery were measured by the New Jersey Department of Health, Public Health and Environmental Laboratories (NJDOH PHEL), a member of the National Biomonitoring Network (NBN) (Ewing, NJ). A CLIA/CLIS certified method that employs inductively coupled plasma mass spectrometry (8900 ICP-QQQ: Agilent Technologies Inc) was used for sample analysis. There were 89 duplicated blinded samples being interspersed, which had less than 5% coefficient of variation (CV). Maternal plasma folate and vitamin B12 levels (from the same blood sample as the RBCs) were measured by a commercial laboratory using chemiluminescent immunoassay (MAGLUMI 2000; Snibe Co, Ltd), with less than 4% CV (Wang et al., 2016). Maternal plasma high-density lipoprotein (HDL) levels from the same blood sample as the RBCs were also measured. The detailed measurement and calculation methods are described in our previous publication (Panjwani, 2020).
Definition of pertinent covariables
This study carefully considered and controlled prenatal and perinatal factors that are known or suspected to affect the risk of neurodevelopmental outcomes, including sex of the child, preterm birth, and maternal race/ethnicity. We also included maternal stress, overweight/obese or Diabetic (OWO/DM), cholesterol levels, and nutritional status (i.e., plasma folate and vitamin B12 levels) based on our prior studies in which established these maternal conditions as risk factors for ASD and/or ADHD (Li, 2016; Okano et al, 2019; Ji et al, 2017; Ji et al, 2018). BBC trained research staff collected maternal demographic and non-clinical variables via standard questionnaire interviews. Medical record abstraction was used to extract clinically related variables for the mothers and children.
Diagnosis of neurodevelopmental disorder (ASD, ADHD, other DD)
We defined four mutually exclusive groups: Autism Spectrum Disorder (ASD), Attention Deficit Hyperactivity Disorder excluding ASD (ADHD), other developmental disabilities excluding ADHD and ASD (other DD), and Neurotypical (NT) children, based on physician diagnoses as documented in their electronic medical records (EMRs) up to June 2018. The EMRs consist of all the primary care and subspecialty visits at BMC since January 2004. International Classification of Diseases - 9 (ICD-9) codes (before October 1, 2015) and ICD-10 (after October 1, 2015) codes were used to document the primary and secondary diagnoses of each visit. “ASD” group is defined as children having ASD-related ICD-9 (299.0–299.91) or ICD-10 (F84.0–F84.9)). “ADHD” group is defined as children having ADHD-related ICD-9 codes (314.0–314.9) or ICD-10 (F90.0–F90.9) codes but excluding ASD-related ICD codes. “Other DD” group is defined as children having ICD-9 (290–319) or ICD-10 (F01–F99) codes related to neurodevelopmental disorders but excluding ASD and ADHD related ICD codes. “NT” consists of children not having any of the above codes related to neurodevelopmental disorders. As shown in supplemental Figure 1 and Table 1, there are 66 dyads in “ASD” group, 216 dyads in “ADHD” group, 617 “other DD” group, and 651 in “NT” group in the dataset for analysis.
Table 1.
Baseline maternal and child characteristics of the study sample across subgroups defined by child developmental diagnoses*.
| NT | ASD | ADHD | Other DD | p-value | ||
|---|---|---|---|---|---|---|
| N | 651 | 66 | 216 | 617 | ||
| Maternal age (%) | <20 | 52 (8.0) | 1 (1.5) | 25 (12.0) | 53 (8.6) | 0.110 |
| 20–35 | 490 (75.3) | 48 (72.7) | 149 (69.0) | 450 (72.9) | ||
| >=35 | 109 (16.7) | 17 (25.8) | 42 (19.4) | 114 (18.5) | ||
| Maternal BMI (mean (SD)) | 26.3 (6.5) | 28.1 (7.2) | 26.6 (6.1) | 27.0 (7.0) | 0.107 | |
| Parity (%) | Nulliparous | 270 (41.5) | 30 (45.5) | 96 (44.4) | 248 (40.2) | 0.653 |
| Multiparous | 381 (58.5) | 36 (54.6) | 120 (55.6) | 369 (59.8) | ||
| Household income (%) | <$30,000 | 222 (34.1) | 35 (53.0) | 78 (36.1) | 262 (42.5) | 0.003 |
| >=$30,000 | 102 (15.7) | 7 (10.6) | 25 (11.6) | 66 (10.7) | ||
| Unknown | 327 (50.2) | 24 (36.4) | 113 (52.3) | 289 (46.8) | ||
| Race/ethnicity (%) | Black | 443 (68.1) | 38 (57.6) | 141 (65.3) | 397 (64.3) | 0.691 |
| White | 31 (4.8) | 4 (6.1) | 15 (6.9) | 42 (6.8) | ||
| Hispanic | 129 (19.8) | 17 (25.8) | 46 (21.3) | 133 (21.6) | ||
| Others | 48 (7.4) | 7 (10.6) | 14 (6.5) | 45 (7.3) | ||
| Marital status (%) | Married | 230 (35.3) | 23 (34.9) | 53 (24.5) | 198 (32.1) | 0.12 |
| Not Married | 415 (63.8) | 43 (65.2) | 162 (75.0) | 415 (67.3) | ||
| Missing | 6 (0.9) | 0 (0.0) | 1 (0.5) | 4 (0.7) | ||
| Maternal education (%) | Below college | 410 (63.3) | 37 (57.8) | 148 (68.5) | 400 (64.9) | 0.355 |
| Above college | 238 (36.7) | 27 (42.2) | 68 (31.5) | 216 (35.1) | ||
| Stressful during pregnancy (%) | No | 286 (43.9) | 25 (31.6) | 72 (34.4) | 223 (36.5) | 0.008 |
| Stressful | 366 (56.1) | 54 (68.4) | 137 (65.6) | 388 (63.5) | ||
| Smoking during pregnancy (%) | Never | 560 (86.0) | 53 (80.3) | 163 (75.5) | 484 (78.4) | 0.001 |
| Quitter | 39 (6.0) | 5 (7.6) | 31 (14.4) | 52 (8.4) | ||
| Continuous | 46 (7.1) | 7 (10.6) | 21 (9.7) | 77 (12.5) | ||
| Missing | 6 (0.9) | 1 (1.5) | 1 (0.5) | 4 (0.7) | ||
| Drinking during pregnancy (%) | No | 596 (92.7) | 57 (89.1) | 188 (89.1) | 561 (91.5) | 0.356 |
| Yes | 47 (7.3) | 7 (10.9) | 23 (10.9) | 52 (8.5) | ||
| Age of the child during the latest visit (mean (SD)) | 7.53 (3.41) | 9.42 (2.93) | 9.69 (2.33) | 8.29 (3.20) | <0.001 | |
| Child’s sex (%) | Female | 372 (57.1) | 19 (28.8) | 60 (27.8) | 314 (50.9) | <0.001 |
| Male | 279 (42.9) | 47 (71.2) | 156 (72.2) | 303 (49.1) | ||
| Delivery type (%) | C/S | 208 (32.0) | 30 (45.5) | 90 (41.7) | 238 (38.6) | 0.008 |
| Vaginal | 443 (68.1) | 36 (54.6) | 126 (58.3) | 379 (61.4) | ||
| Preterm birth (%) | No | 527 (81.0) | 44 (66.7) | 154 (71.3) | 439 (71.2) | <0.001 |
| Yes | 124 (19.1) | 22 (33.3) | 62 (28.7) | 178 (28.9) | ||
| Low birthweight (%) | No | 536 (82.3) | 45 (68.2) | 159 (73.6) | 450 (72.9) | <0.001 |
| Yes | 115 (17.7) | 21 (31.8) | 57 (26.4) | 167 (27.1) | ||
| Selenium levels in μg/L (mean (SD)) | 286.16 (56.83) | 302.49 (85.16) | 291.75 (59.42) | 289.61 (64.07) | 0.17 | |
| Selenium quartiles (%) | Q1 | 163 (25.0) | 14 (21.2) | 47 (21.8) | 164 (26.6) | 0.617 |
| Q2 | 176 (27.0) | 14 (21.2) | 63 (29.2) | 146 (23.7) | ||
| Q3 | 165 (25.4) | 22 (33.3) | 55 (25.5) | 154 (25.0) | ||
| Q4 | 147 (22.6) | 16 (24.2) | 51 (23.6) | 153 (24.8) | ||
ASD: autism spectrum disorder; ADHD: Attention deficit hyperactivity disorder excluding ASD; other DD: other developmental disabilities excluding ASD and ADHD.
Note: p-value was calculated by ANOVA and Chi-square test for continuous and categorical variables respectively.
Statistical methods
The maternal and child characteristics of the study children by the four (ASD, ADHD, other DD, and NT) groups were compared using Pearson’s chi-squared test (or Fisher’s exact test for small cells) and ANOVA test for categorical and continuous variables, respectively. The distribution of maternal selenium levels was compared across the four groups. The probabilities of diagnosis for ASD, ADHD, and other DD were compared across maternal selenium levels. Next, maternal selenium levels were analyzed in following three ways: converted into quartiles, binary (below or above median), and continuous variable divided by inter-quartile range. Maternal plasma folate, vitamin B12, and HDL were categorized into binary variables (below vs above median). Maternal overweight/obesity and maternal diabetes were combined into a new binary variable (OWO/DM): no obesity and no diabetes vs. any obesity and/or diabetes to represent the maternal metabolic condition. The reason we combined OWO and DM in the analysis is that obesity and diabetes are major inflammatory metabolic disorders marked by increased oxidative stress. In an earlier report in the cohort, we showed that maternal diabetes and obesity were associated with an increased risk of ASD (Li et al., 2016). Additional reason for such a combination is due to the correlation between the two (potentially co-linear) and sample size constraint.
Multiple imputation by chained equations (MICE) method was used to impute missing data for sociodemographic characteristics (less than 4%) via mice package in R (Van Buuren & Groothuis-Oudshoorn, 2011). Adjusted logistic regression models were used to examine the associations of maternal RBC selenium levels with child’s risk of ASD, ADHD and other DD, respectively, using NT children as the common reference group. The model adjusted for maternal age at delivery, maternal race/ethnicity, maternal education, marital status, stress during pregnancy, smoking before or during pregnancy, alcohol drinking before or during pregnancy, parity, child’s sex, delivery type, preterm birth, and birthweight. We further tested the combined associations for maternal obesity/diabetes and red blood cell selenium levels on the outcomes. When maternal BMI data was missing in a small number of participants (4.7%), we assigned those to the non-obese category.
As a sensitivity analysis, we performed stratified analyses by each stratum of important covariates, including child’s sex, maternal race/ethnicity, preterm birth, maternal OWO/DM (OWO is defined as BMI≥30 kg/m2) (Li et al. 2016), maternal HDL, and maternal folate and vitamin B12. Within each stratum of important covariates, multiple logistic regression models were conducted between selenium/IQR (divided by total sample inter-quartile range) and each neurodevelopmental outcome groups, adjusting for the same covariates except for the stratified one. (Mean: 289.0 μg/L, Median: 278 μg/L, IQR: 246, 318 μg/L). Furthermore, we tested the combined associations of maternal OWO/DM and RBC selenium levels with the outcomes, adjusting for maternal folate level; testing the combined associations of maternal folate and RBC selenium levels with the outcomes; testing the combined associations of maternal Vitamin B12 and RBC selenium levels on the outcomes; and testing the combined associations of maternal OWO/DM and RBC selenium levels with the risk of any developmental disabilities. R 3.4.3 software was used to perform all analyses (R Core Team, 2018).
Results
Table 1 shows the baseline characteristics of 1,550 mother-infant pairs from the BBC. ANOVA and Chi-square tests show that there are statistically significant differences among child developmental diagnoses subgroups in terms of household income, marital status, stress status during pregnancy, smoking status during pregnancy, child age of latest clinical visit, child’s sex, delivery type, preterm birth, and low birthweight.
Maternal RBC selenium levels and child risk of ASD, ADHD, and other DD
The mean maternal selenium levels for NT, ASD, ADHD, and other DD were 286.16 μg/L, 302.49 μg /L, 291.75 μg /L, and 289.61 μg/L, respectively. The median is 278 μg/L (inter-quartile range: 246, 318 μg/L). The distribution curve of maternal selenium levels in ASD group displayed a slight rightward shift compared to that of NT group (Figure 1). The probability of ASD diagnosis decreased at lower maternal selenium RBC levels but increased once exceeding a particular threshold (Figure 2). There was a positive, dose-dependent association between maternal RBC selenium levels and child risk of developing ASD, with an adjusted odds ratio of 1.49 (95% CI: 1.09, 2.02, p=0.012) per IQR increase in selenium (Table 2). There was a similar positive trend of association for ADHD with an adjusted odds ratio of 1.29 (95% CI: 1.04, 1.56, p=0.020) per IQR increase (Table 2). This trend was not seen for other DD.
Figure 1.
Distribution of maternal red blood cell (RBC) selenium levels, stratified by child developmental diagnoses, including Autism Spectrum Disorder (ASD), Attention Deficit Hyperactivity Disorder (ADHD), other Developmental Disabilities (other DD), and Neurotypical (NT).
Figure 2.
Smoothing plot of probability of developmental diagnoses by maternal red blood cell (RBC) selenium levels, including Autism Spectrum Disorder (ASD), Attention Deficit Hyperactivity Disorder (ADHD), other Developmental Disabilities (other DD), and Neurotypical (NT)
Table 2.
Individual and combined associations of maternal overweight/obesity or diabetes (OWO/DM) and red blood cell (RBC) selenium levels with child risk of developmental disabilities*
| NT, N=651 | ASD, N=66 | ADHD, N=216 | Other DD, N=617 | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Selenium | Ob or DM | Total | N (%) | N (%) | aOR# | 95% CI | P-value | N (%) | aOR# | 95% CI | P-value | N (%) | aOR# | 95% CI | P-value | |
| Selenium/IQR** | Ref | 1.49 | [1.09, 2.02] | 0.012 | 1.29 | [1.04, 1.56] | 0.020 | 1.12 | [0.97, 1.29] | 0.120 | ||||||
| Quartiles | Q1 | 388 | 163 (42.0) | 14 (3.6) | Ref | 47 (12.1) | Ref | 164 (42.3) | Ref | |||||||
| Q2 | 399 | 176 (44.1) | 14 (3.5) | 1.07 | [0.46, 2.47] | 0.876 | 63 (15.8) | 1.35 | [0.83, 2.18] | 0.205 | 146 (36.6) | 0.89 | [0.64, 1.23] | 0.484 | ||
| Q3 | 396 | 165 (41.7) | 22 (5.6) | 1.88 | [0.86, 4.14] | 0.115 | 55 (13.9) | 1.20 | [0.73, 1.97] | 0.308 | 154 (38.9) | 1.06 | [0.76, 1.47] | 0.739 | ||
| Q4 | 367 | 147 (40.1) | 16 (4.4) | 1.67 | [0.71, 3.94] | 0.240 | 51 (13.9) | 1.54 | [0.93, 2.57] | 0.144 | 153 (41.7) | 1.15 | [0.82, 1.60] | 0.420 | ||
| Binary | Below median | 758 | 327 (43.1) | 28 (3.7) | Ref | 103 (13.6) | Ref | 300 (39.6) | Ref | |||||||
| Above median | 792 | 324 (40.9) | 38 (4.8) | 1.62 | [0.90, 2.90] | 0.105 | 113 (14.3) | 1.22 | [0.87, 1.72] | 0.255 | 317 (40.0) | 1.17 | [0.93, 1.49] | 0.182 | ||
| None | 1078 | 475 (44.1) | 40 (3.7) | Ref | 151 (14.0) | Ref | 412 (38.2) | Ref | ||||||||
| Any | 472 | 176 (37.3) | 26 (5.5) | 1.73 | [0.97, 3.09] | 0.06 | 65 (13.8) | 1.16 | [0.80, 1.69] | 0.423 | 205 (43.4) | 1.34 | [1.04, 1.74] | 0.023 | ||
| Combined | Below median | None | 522 | 235 (45.0) | 17 (3.3) | Ref | 71 (13.6) | Ref | 199 (38.1) | Ref | ||||||
| Below median | Any | 236 | 92 (39.0) | 11 (4.7) | 1.36 | [0.57, 3.22] | 0.484 | 32 (13.6) | 1.24 | [0.73, 2.09] | 0.424 | 101 (42.8) | 1.30 | [0.91, 1.87] | 0.141 | |
| Above median | None | 556 | 240 (43.2) | 23 (4.1) | 1.39 | [0.67, 2.87] | 0.369 | 80 (14.4) | 1.27 | [0.85, 1.89] | 0.253 | 213 (38.3) | 1.16 | [0.88, 1.53] | 0.301 | |
| Above median | Any | 236 | 84 (35.6) | 15 (6.4) | 3.10 | [1.35, 7.15] | 0.008 | 33 (14.0) | 1.41 | [0.82, 2.43] | 0.212 | 104 (44.1) | 1.61 | [1.12, 2.32] | 0.011 | |
| Test of interaction between Ob/DM and Selenium binary | 0.400 | 0.781 | 0.812 | |||||||||||||
ASD: autism spectrum disorder; ADHD: Attention deficit hyperactivity disorder excluding ASD; other DD: other developmental disabilities excluding ASD and ADHD.
IQR: Inter-quartile range = 72 (246, 318 μg/L)
Multiple logistic regression adjusted for: maternal age, race, maternal education, marital status, smoking, alcohol drinking, stress during pregnancy, parity, preterm birth, low birthweight, child’s sex, and delivery type.
Overall, our findings remained consistent even after performing stratified analyses as shown in Table 3. The association between maternal selenium levels (presented as per IQR increase) and risk of ASD was significant among black mothers (p =0.014), term birth (p=0.017), having OWO/DM (p=0.016), below median maternal HDL (p =0.002), above median maternal folate (p-value=0.034), and above median maternal B12 groups (p=0.013). The association between maternal selenium levels and risk of ADHD were significant among female child (p=0.029), non-black mothers (p =0.009), absence of maternal OWO/DM (p =0.023), above median maternal HDL (p =0.007) and below median maternal folate (p=0.044) groups. The association between maternal selenium levels and risk of other DD diagnosis were significant among below median maternal HDL (p =0.016). Supplemental Table 1 replicated Table 2 while further adjusting for maternal folate level, producing comparable results.
Table 3.
Sensitivity analysis: Associations of maternal red blood cell (RBC) selenium levels with child risk of developmental disabilities in subgroups*.
| ASD | ADHD | Other DD | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Stratified by | aOR | 95% CI | P-value | aOR | 95% CI | P-value | aOR | 95% CI | P-value | |
| Child’s Sex | ||||||||||
| Female | Selenium/IQR | 1.40 | [0.84, 2.34] | 0.191 | 1.42 | [1.04, 1.95] | 0.029 | 1.11 | [0.91, 1.34] | 0.296 |
| Male | Selenium/IQR | 1.49 | [0.99, 2.24] | 0.074 | 1.23 | [0.91, 1.65] | 0.180 | 1.14 | [0.92, 1.42] | 0.235 |
| Race/ethnicity | ||||||||||
| Black | Selenium/IQR | 1.56 | [1.09, 2.23] | 0.014 | 1.19 | [0.92, 1.53] | 0.187 | 1.09 | [0.92, 1.30] | 0.299 |
| Non-Black | Selenium/IQR | 1.18 | [0.59, 2.37] | 0.635 | 1.76 | [1.15, 2.69] | 0.009 | 1.20 | [0.92, 1.57] | 0.177 |
| Preterm birth | ||||||||||
| No | Selenium/IQR | 1.56 | [1.10, 2.23] | 0.017 | 1.22 | [0.94, 1.58] | 0.133 | 1.07 | [0.90, 1.27] | 0.425 |
| Yes | Selenium/IQR | 1.51 | [0.75, 3.05] | 0.253 | 1.39 | [0.93, 2.08] | 0.107 | 1.30 | [0.97, 1.74] | 0.082 |
| Obesity/DM | ||||||||||
| No | Selenium/IQR | 1.44 | [0.95, 2.17] | 0.082 | 1.33 | [1.04, 1.69] | 0.023 | 1.06 | [0.89, 1.26] | 0.506 |
| Yes | Selenium/IQR | 2.03 | [1.14, 3.61] | 0.016 | 1.25 | [0.78, 2.00] | 0.353 | 1.30 | [0.98, 1.71] | 0.065 |
| Maternal HDL | ||||||||||
| Below median | Selenium/IQR | 2.04 | [1.31, 3.16] | 0.002 | 1.32 | [0.94, 1.86] | 0.106 | 1.34 | [1.06, 1.70] | 0.016 |
| Above median | Selenium/IQR | 1.01 | [0.50, 2.04] | 0.980 | 1.66 | [1.15, 2.41] | 0.007 | 1.08 | [0.87, 1.35] | 0.474 |
| Maternal folate | ||||||||||
| Below median | Selenium/IQR | 1.39 | [0.77, 2.49] | 0.270 | 1.38 | [1.01, 1.88] | 0.044 | 1.17 | [0.94, 1.45] | 0.169 |
| Above median | Selenium/IQR | 1.72 | [1.04, 2.83] | 0.034 | 1.35 | [0.89, 2.05] | 0.153 | 1.24 | [0.96, 1.60] | 0.093 |
| Maternal vitamin B12 | ||||||||||
| Below median | Selenium/IQR | 0.92 | [0.48, 1.76] | 0.451 | 1.33 | [0.94, 1.87] | 0.141 | 1.21 | [0.95, 1.54] | 0.118 |
| Above median | Selenium/IQR | 1.73 | [1.12, 2.68] | 0.013 | 1.30 | [0.90, 1.87] | 0.156 | 1.11 | [0.89, 1.40] | 0.357 |
ASD: autism spectrum disorder; ADHD: Attention deficit hyperactivity disorder excluding ASD; other DD: other developmental disabilities excluding ASD and ADHD.
Multiple logistic Regression adjusted for: maternal age, race (dropped when stratified), maternal education, marital status, smoking, alcohol drinking, stress during pregnancy, parity, maternal ob/DM, preterm birth (dropped when stratified), low birthweight, child’s sex (dropped when stratified), and delivery type.
P>0.05 for test of interaction between selenium and each stratified variable.
Combined associations between maternal RBC selenium levels and OWO/DM and child risk of ASD, ADHD, and other DD
Compared to mothers with below median selenium and no OWO/DM condition, the mothers with elevated selenium levels and OWO/DM condition have 210% increased odds of having ASD diagnosis for their children (OR: 3.10; 95% CI: 1.35, 7.15, p=0.008) and 61% increased odds of having other DD diagnosis for their children (OR: 1.61; 95% CI: 1.12, 2.32, p=0.011),while such trend was not observed in ADHD (Table 2), and such findings were consistent when further adjusted for maternal folate level (Supplement Table 1). The test of interaction between OWO/DM and selenium binary was not significant for any of the neurodevelopmental diagnoses (p>0.05) (Table 2).
Of note, the adjusted odds ratio of any developmental disorder as a whole (including ASD, ADHD, and other DD) with maternal OWO/DM and above median selenium levels was 1.66 (95% CI: 1.18, 2.33, p=0.004) (Supplemental Table 4). The test of interaction between OWO/DM and selenium binary for any of the neurodevelopmental diagnoses was not significant (p>0.05).
Combined associations between maternal RBC selenium levels and nutritional status and child risk of ASD, ADHD, and other DD
Folate
Among mothers with above median folate levels, increasing selenium levels by one IQR was significantly associated with 52% increased odds of ASD diagnosis (aOR:1.52; 95% CI: 1.11, 2.07, p=0.008). The risk of ASD was not statistically significant in the presence of both above median maternal serum folate and above median maternal RBC selenium levels (aOR 1.98: 95% CI: 0.86, 4.54, p=0.108). Similarly, there were no significant combined associations observed for ADHD and other DD. The test for interaction between maternal folate and selenium binary was not significant for any neurodevelopmental diagnosis (p>0.05) (Supplemental Table 2).
Vitamin B12
Among mothers with above median vitamin B12 levels, increasing selenium levels by one IQR was significantly associated with a 52% increased odd of having ASD diagnosis (aOR: 1.52; 95% CI: 1.11, 2.07, p=0.008). No significant combined associations were seen for ASD. The only significant combined association observed was an elevated risk of other DD in children both from mothers with below median vitamin B12 and above median selenium levels (aOR 1.50; 95% CI: 1.04, 2.16, p=0.029). The test for interaction between maternal vitamin B12 and selenium binary was not significant for any of the neurodevelopmental diagnoses (p>0.05) (Supplemental Table 3).
Discussion
To our knowledge, this is the first prospective cohort study in the US to examine the role of prenatal selenium exposure in child risk of ASD. While ASD is our primary outcome of interest, we have also included ADHD and other neurodevelopmental disorders as parallel analyses, given that these other conditions often co-exist with ASD. We also explored whether other known or suspected risk factors of ASD could confound or modify the associations, including child sex, preterm birth, maternal race, OWO/DM, maternal HDL, maternal folate, and maternal B12.
We showed that maternal selenium exposure contributed to ASD risk in a dose-dependent fashion, with children born from mothers with the highest selenium levels in the cohort having nearly two-fold risk of developing ASD, compared to children born from mothers whose selenium levels were in the lowest quartile. Such dose-dependent effect was further amplified when additional maternal characteristics including OWO/DM and elevated folate, or vitamin B12 levels, were considered. Our data not only support our hypothesis that excess selenium exposure may increase child ASD risk, but also suggest there are additional factors which may further potentiate selenium’s neurodevelopmental influence.
The biological mechanisms underlying the selenium-ASD association remain to be understood. While limited, previous studies have provided some possible explanations. Misra et al reported increased oxidative stress and decreased viability in trout hepatocytes with increasing selenium concentrations (Misra & Niyogi, 2009), which suggests that while selenium is essential for modulating oxidative stress, it may be cytotoxic at excess amounts. Hagmeyer at el showed that excess selenium levels correlated to a decrease in the number of neuronal synapses, as well as an exponential increase in rat hippocampal cell death (Hagmeyer, Mangus, Boeckers & Grabrucker, 2015). In addition, Raymond et al reported diminished selenium-dependent antioxidant levels in children with ASD compared to those without ASD (Jory & McGinnis, 2008).
Our findings have implications for maternal and child health. In the US, we currently lack guidelines for selenium supplementation in pregnant mothers, and there is limited data on the long-term neurodevelopmental outcomes of prenatal selenium exposure. According to the Institute of Medicine of the National Academies Food and Nutrition Board (FNB), the recommended daily intake of selenium for an average adult is 55mcg and 60mcg for pregnant women. Tolerable upper limit for an average adult, including pregnant women, is 400mcg per day (National Institutes of Health [NIH], 2020). While selenium is found in multiple food sources including grain products (13mcg of selenium/serving), meat (18mcg of selenium/serving), and dairy (8mcg of selenium/serving) (National Institutes of Health [NIH], 2020), it is also found in prenatal vitamins, or in separate supplement pills containing up to 200mcg of selenium per serving (Stranges et al., 2006). Pregnant women may underestimate their total daily selenium levels not taking into account both food sources and supplementation. Results from our study, if further confirmed, suggest that screening women for selenium levels during pregnancy may help identify mothers whose selenium levels are not optimal for their child neurodevelopment outcomes, so that health care providers may guide appropriate intake to ensure selenium levels remain within the optimal range.
In addition, it would be prudent to consider other potential risk factors for ASD to identify optimal selenium levels in the presence of these risk factors, such as maternal OWO/DM and folate/vitamin B12 levels. At the biochemical level, folate, vitamin B12, and selenium participate in a series of reactions necessary for gene regulation. These reactions belong to the methionine-homocysteine cycle, in which folate, vitamin B12, and selenium act as co-factors for enzymes involved in DNA methylation. The methionine-homocysteine cycle also produces glutathione peroxidase, a selenium-dependent enzyme that protects cells against oxidative damage (Joseph & Loscalzo, 2013). While all three nutrients are vital to cellular function, there is evidence that they can lead to dysfunction of gene regulation and increased oxidative stress at supratherapeutic levels. For instance, Barua et al reported that excess folic acid in rat cerebellum was linked to altered gene expression of several genes reported to impact ASD pathogenesis (Barua, Kuizon, Chadman, Brown & Junaid, 2015). In addition, we have previously shown that there is a “U” shaped relationship between maternal folate levels and risk of child ASD. At very high concentrations, folate and vitamin B12 were each associated with more than a 2-fold increase in ASD risk (Raghavan et al., 2018). The pathophysiology of combined associations between maternal selenium and folate/vitamin B12 with ASD risk remains unclear, but it is biologically plausible that excess concentrations of these cofactors may lead to dysregulation of the redox/methylation reactions necessary for brain development.
Strengths and Limitations
A major strength of our study includes assessing the association between prenatal selenium exposure and neurodevelopmental outcomes in a large, prospective cohort study. We assessed prenatal selenium exposure by using maternal RBC selenium levels as a proxy of fetal exposure in third trimester. In a separate study, we have previously shown that maternal RBC selenium levels better reflect mother-to-fetus transfer of selenium compared to plasma (Chen et al., 2014). Selenium lab analysis was conducted by an NBN member laboratory. The CLIA-certified method employed assures data are of high quality and are comparable to other data generated by the NBN laboratories. This harmonization also reduces uncertainty for assessment of health effects related to maternal selenium levels. An added advantage of maternal selenium assessment is that it offers an opportunity for risk assessment and intervention during the critical fetal developmental period.
Our study has following limitations. Maternal RBC selenium levels were measured only once, which does not account for variability in selenium levels at conception or during early gestational age. We did not measure selenium intake in our cohort population, which is a major source of selenium and a key contributor to selenium metabolism in the body. Our ASD cases are based on EMR data (ICD-9/10 codes) and not based on gold standard assessment. It is also important to note that the exact pathophysiology of ASD remains unknown, and that there may be genetic predispositions to ASD. We did not consider all nutritional biomarkers that have been linked to redox/methylation reactions related to ASD etiology. For instance, vitamin B6 is a cofactor of the glutathione pathway (Main, Angley, O’Doherty, Thomas & Fenech, 2012), which protects the body from oxidative stress. Our cohort consisted of an urban, low-income minority population in the US; cautions are needed when generalizing our findings to other populations with different characteristics. Due to the variability of plasma/serum selenium concentrations in pregnant women depending on their geographical location and gestational age, there is no consensus for what “normal” level of plasma selenium (or RBC selenium) is in pregnant women, which may present challenges with providing specific anticipatory guidance. Multiple comparison could be a concern with many models and outcomes. Previous epidemiologic studies have identified numerous early life risk factors for risk of ASD, such as maternal obesity (Li et al., 2016), diabetes (Buchmayer et al., 2009), intrauterine infection/inflammation (IUI) (Brown et al., 2014), preeclampsia (Mann, McDermott, Bao, Hardin & Gregg, 2010), prenatal stress (Beversdorf et al., 2005), prenatal use of psychotropic drugs, prematurity (Johnson et al., 2010), or low birthweight (Muratore et al., 2013), male gender (Center for Disease Control and Prevention [CDC], 2018; Van Naarden Braun et al., 2015; Jeste, 2015), and genetic factors (Miles, 2011; Currenti, 2010). Together with our current evidence, ASD is highly likely to have a multifactorial etiology. Given this multifactorial nature, future studies should apply advanced, sophisticated methodologies to evaluate the impact of multiple risk factors and environmental exposures simultaneously. Finally, we have focused on ASD as our primary outcome of interest but it is important to note that there is observed clinical overlap between ASD and ADHD, and that ASD is associated with other neurodevelopmental disorders not just limited to ADHD. Additional larger studies are needed to better delineate the similarity and difference in the association between selenium and ASD, ADHD, and other DD, to gain new insight into their shared early life factors and biological mechanisms.
Conclusion
In this prospective birth cohort study, we observed a significant dose-dependent association between maternal RBC selenium levels and child risk of developing ASD or ADHD. Our study raises more questions than we can answer. Given that selenium is a component of multivitamins and environmental selenium varies greatly by geographic regions, we hope that our study will stimulate more research to clarify how we can maximize selenium’s health benefits and minimize its harm on pregnancy and long-term health outcomes. If confirmed, optimizing maternal prenatal selenium levels may be a novel intervention to help prevent neurodevelopmental disorders in children.
Supplementary Material
Acknowledgments:
The authors thank Linda Rosen of the Boston University Clinical Data Warehouse for assistance in obtaining relevant clinical information; the Clinical Data Warehouse service is supported by Boston University Clinical and Translational Institute and the National Institutes of Health Clinical and Translational Science Award (grant U54-TR001012).
Funding/Support:
This study is supported in part by the Health Resources and Services Administration (HRSA) of the U.S. Department of Health and Human Services (HHS) under grant number R40MC27443, Autism Field-initiated Innovative Research Studies Program; and grant number UJ2MC31074, Autism Single Investigator Innovation Program. The Boston Birth Cohort (the parent study) is supported in part by the National Institutes of Health (NIH) grants (R01HD086013, 2R01HD041702, R01HD098232, R01 ES031272, and 1R01ES031521). This study was possible in part due to funding for programmatic analytical capacity and capability through Cooperative Agreement #CDC-RFA-EH14-140203 between the New Jersey Department of Health (NJDOH) Public Health and Environmental Laboratories (PHEL) and the Centers for Disease Control and Prevention (CDC) States Biomonitoring Grant Program.
Abbreviations:
- ADHD
Attention deficit hyperactivity disorder
- ASD
Autism Spectrum Disorder
- BBC
Boston Birth Cohort
- RBC
Red blood cells
- ICD
International Classification of Diseases
- Se
Selenium
- NJDOH PHEL
New Jersey Department of Health, Public Health and Environmental Laboratories
Footnotes
Publisher's Disclaimer: Disclaimer: This information or content and conclusions are those of the authors and should not be construed as the official position or policy of, nor should any endorsements be inferred by, the HRSA, HHS, the NJDOH, the CDC, or the U.S. Government. The funding agencies were not involved in the collection, analysis, or interpretation of data; in the writing of the report; or in the decision to submit the article for publication.
Conflict of Interest: The authors declare no other conflict of interest
References
- Barua S, Kuizon S, Chadman KK, Brown WT, & Junaid MA (2015). Microarray analysis reveals higher gestational folic Acid alters expression of genes in the cerebellum of mice offspring-a pilot study. Brain sciences, 5(1), 14–31. 10.3390/brainsci5010014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Beversdorf DQ, Manning SE, Hillier A, Anderson SL, Nordgren RE, Walters SE, Nagaraja HN, Cooley WC, Gaelic SE, & Bauman ML (2005). Timing of prenatal stressors and autism. Journal of autism and developmental disorders, 35(4), 471–478. 10.1007/s10803-005-5037-8 [DOI] [PubMed] [Google Scholar]
- Bjørklund G, Meguid NA, El-Bana MA, Tinkov AA, Saad K, Dadar M, Hemimi M, Skalny AV, Hosnedlová B, Kizek R, Osredkar J, Urbina MA, Fabjan T, El-Houfey AA, Kałużna-Czaplińska J, Gątarek P, & Chirumbolo S (2020). Oxidative Stress in Autism Spectrum Disorder. Molecular neurobiology, 57(5), 2314–2332. 10.1007/s12035-019-01742-2 [DOI] [PubMed] [Google Scholar]
- Brown AS, Sourander A, Hinkka-Yli-Salomäki S, McKeague IW, Sundvall J, & Surcel HM (2014). Elevated maternal C-reactive protein and autism in a national birth cohort. Molecular psychiatry, 19(2), 259–264. 10.1038/mp.2012.197 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Buchmayer S, Johansson S, Johansson A, Hultman CM, Sparén P, & Cnattingius S (2009). Can association between preterm birth and autism be explained by maternal or neonatal morbidity?. Pediatrics, 124(5), e817–e825. 10.1542/peds.2008-3582 [DOI] [PubMed] [Google Scholar]
- Chen Z, Myers R, Wei T, Bind E, Kassim P, Wang G, Ji Y, Hong X, Caruso D, Bartell T, Gong Y, Strickland P, Navas-Acien A, Guallar E, & Wang X (2014). Placental transfer and concentrations of cadmium, mercury, lead, and selenium in mothers, newborns, and young children. Journal of exposure science & environmental epidemiology, 24(5), 537–544. 10.1038/jes.2014.26 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Currenti SA (2010). Understanding and determining the etiology of autism. Cellular and molecular neurobiology, 30(2), 161–171. 10.1007/s10571-009-9453-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- De Palma G, Catalani S, Franco A, Brighenti M, & Apostoli P (2012). Lack of correlation between metallic elements analyzed in hair by ICP-MS and autism. Journal of autism and developmental disorders, 42(3), 342–353. 10.1007/s10803-011-1245-6 [DOI] [PubMed] [Google Scholar]
- Fact Sheet for Health Professionals (2020). In Selenium. Retrieved from National Institues of Health: Office of Dietary Supplements. ods.od.nih.gov/factsheets/Selenium-HealthProfessional/ [Google Scholar]
- Hagmeyer S, Mangus K, Boeckers TM, & Grabrucker AM (2015). Effects of trace metal profiles characteristic for autism on synapses in cultured neurons. Neural plasticity, 2015, 985083. 10.1155/2015/985083 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Jeste SS (2015). Neurodevelopmental behavioral and cognitive disorders. Continuum (Minneapolis, Minn.), 21(3 Behavioral Neurology and Neuropsychiatry), 690–714. 10.1212/01.CON.0000466661.89908.3c [DOI] [PubMed] [Google Scholar]
- Ji Y, Hong X, Wang G, Chatterjee N, Riley AW, Lee LC, Surkan P, Bartell TR, Zuckerman B, Wang X. A prospective birth cohort study on early childhood lead levels and attention deficit hyperactivity disorder: new insight on sex differences. J of Pediatrics. 2018. DOI: 10.1016/j.jpeds.2018.03.076 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ji Y, Riley AW, Lee LC, Volk H, Hong X, Wang G, Angomas R, Stivers T, Wahl A, Ji H, Bartell TR, Burd I, Paige D, Fallin MD, Zuckerman B, Wang X. A prospective birth cohort study on maternal cholesterol levels and offspring attention deficit hyperactivity disorder: new insight on sex differences. Brain Sci. 2017. December 23;8(1). pii: E3. doi: 10.3390/brainsci8010003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Johnson S, Hollis C, Kochhar P, Hennessy E, Wolke D, & Marlow N (2010). Autism spectrum disorders in extremely preterm children. The Journal of pediatrics, 156(4), 525–31. e2. 10.1016/j.jpeds.2009.10.041 [DOI] [PubMed] [Google Scholar]
- Joseph J, & Loscalzo J (2013). Selenistasis: epistatic effects of selenium on cardiovascular phenotype. Nutrients, 5(2), 340–358. 10.3390/nu5020340 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kiełczykowska M, Kocot J, Paździor M, & Musik I (2018). Selenium - a fascinating antioxidant of protective properties. Advances in clinical and experimental medicine : official organ Wroclaw Medical University, 27(2), 245–255. 10.17219/acem/67222 [DOI] [PubMed] [Google Scholar]
- Kieliszek M (2019). Selenium⁻Fascinating Microelement, Properties and Sources in Food. Molecules (Basel, Switzerland), 24(7), 1298. 10.3390/molecules24071298 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kumar R, Tsai HJ, Hong X, Liu X, Wang G, Pearson C, Ortiz K, Fu M, Pongracic JA, Bauchner H, & Wang X (2011). Race, ancestry, and development of food-allergen sensitization in early childhood. Pediatrics, 128(4), e821–e829. 10.1542/peds.2011-0691 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lakshmi Priya MD, Geetha A Level of Trace Elements (Copper, Zinc, Magnesium and Selenium) and Toxic Elements (Lead and Mercury) in the Hair and Nail of Children with Autism. Biol Trace Elem Res 142, 148–158 (2011). 10.1007/s12011-010-8766-2 [DOI] [PubMed] [Google Scholar]
- Okano L, Ji Y, Riley AW, Wang X. Maternal psychosocial stress and children’s ADHD diagnosis: a prospective birth cohort study. J Psychosom Obstet Gynaecol. 2019. September;40(3):217–225. doi: 10.1080/0167482X.2018.1468434. Epub 2018 May 23. (available on 2020-09-01) [DOI] [PMC free article] [PubMed] [Google Scholar]
- Li M, Fallin MD, Riley A, Landa R, Walker SO, Silverstein M, Caruso D, Pearson C, Kiang S, Dahm JL, Hong X, Wang G, Wang MC, Zuckerman B, & Wang X (2016). The Association of Maternal Obesity and Diabetes With Autism and Other Developmental Disabilities. Pediatrics, 137(2), e20152206. 10.1542/peds.2015-2206 [DOI] [PMC free article] [PubMed] [Google Scholar]
- MacFarquhar JK, Broussard DL, Melstrom P, Hutchinson R, Wolkin A, Martin C, Burk RF, Dunn JR, Green AL, Hammond R, Schaffner W, & Jones TF (2010). Acute selenium toxicity associated with a dietary supplement. Archives of internal medicine, 170(3), 256–261. 10.1001/archinternmed.2009.495 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Maenner MJ, Shaw KA, Baio J, et al. (2016). Prevalence of Autism Spectrum Disorder Among Children Aged 8 Years—Autism and Developmental Disabilities Monitoring Network, 11 Sites, MMWR Surveill Summ 2020; 69 (SS-4), 1–12. 10.15585/mmwr.ss6904a1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Main PA, Angley MT, O’Doherty CE, Thomas P, & Fenech M (2012). The potential role of the antioxidant and detoxification properties of glutathione in autism spectrum disorders: a systematic review and meta-analysis. Nutrition & metabolism, 9, 35. 10.1186/1743-7075-9-35 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Mann JR, McDermott S, Bao H, Hardin J, & Gregg A (2010). Pre-eclampsia, birth weight, and autism spectrum disorders. Journal of autism and developmental disorders, 40(5), 548–554. 10.1007/s10803-009-0903-4 [DOI] [PubMed] [Google Scholar]
- Miles JH (2011). Autism spectrum disorders--a genetics review. Genetics in medicine : official journal of the American College of Medical Genetics, 13(4), 278–294. 10.1097/GIM.0b013e3181ff67ba [DOI] [PubMed] [Google Scholar]
- Misra S, & Niyogi S (2009). Selenite causes cytotoxicity in rainbow trout (Oncorhynchus mykiss) hepatocytes by inducing oxidative stress. Toxicology in vitro : an international journal published in association with BIBRA, 23(7), 1249–1258. 10.1016/j.tiv.2009.07.031 [DOI] [PubMed] [Google Scholar]
- Muratore CR, Hodgson NW, Trivedi MS, Abdolmaleky HM, Persico AM, Lintas C, De la Monte S, & Deth RC (2013). Age-dependent decrease and alternative splicing of methionine synthase mRNA in human cerebral cortex and an accelerated decrease in autism. PloS one, 8(2), e56927. 10.1371/journal.pone.0056927 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Pangrazzi L, Balasco L, & Bozzi Y (2020). Oxidative Stress and Immune System Dysfunction in Autism Spectrum Disorders. International journal of molecular sciences, 21(9), 3293. 10.3390/ijms21093293 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Panjwani AA, Ji Y, Fahey JW, Palmer A, Wang G, Hong X, Zuckerman B, & Wang X (2020). Maternal Dyslipidemia, Plasma Branched-Chain Amino Acids, and the Risk of Child Autism Spectrum Disorder: Evidence of Sex Difference. Journal of autism and developmental disorders, 50(2), 540–550. 10.1007/s10803-019-04264-x [DOI] [PubMed] [Google Scholar]
- Raghavan R, Riley AW, Volk H, Caruso D, Hironaka L, Sices L, Hong X, Wang G, Ji Y, Brucato M, Wahl A, Stivers T, Pearson C, Zuckerman B, Stuart EA, Landa R, Fallin MD, & Wang X (2018). Maternal Multivitamin Intake, Plasma Folate and Vitamin B12 Levels and Autism Spectrum Disorder Risk in Offspring. Paediatric and perinatal epidemiology, 32(1), 100–111. 10.1111/ppe.12414 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Rai D, Lee BK, Dalman C, Golding J, Lewis G, & Magnusson C (2013). Parental depression, maternal antidepressant use during pregnancy, and risk of autism spectrum disorders: population based case-control study. BMJ (Clinical research ed.), 346, f2059. 10.1136/bmj.f2059 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Raymond LJ, Deth RC, & Ralston NV (2014). Potential Role of Selenoenzymes and Antioxidant Metabolism in relation to Autism Etiology and Pathology. Autism research and treatment, 2014, 164938. 10.1155/2014/164938 [DOI] [PMC free article] [PubMed] [Google Scholar]
- R Core Team (2018). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Available online at https://www.R-project.org/ [Google Scholar]
- Stranges S, Marshall JR, Trevisan M, Natarajan R, Donahue RP, Combs GF, Farinaro E, Clark LC, & Reid ME (2006). Effects of selenium supplementation on cardiovascular disease incidence and mortality: secondary analyses in a randomized clinical trial. American journal of epidemiology, 163(8), 694–699. 10.1093/aje/kwj097 [DOI] [PubMed] [Google Scholar]
- Van Buuren S, & Groothuis-Oudshoorn K (2011). mice: Multivariate Imputation by Chained Equations in R. Journal of Statistical Software, 45(3), 1–67. doi: 10.18637/jss.v045.i03 [DOI] [Google Scholar]
- Van Naarden Braun K, Christensen D, Doernberg N, Schieve L, Rice C, Wiggins L, Schendel D, & Yeargin-Allsopp M (2015). Trends in the prevalence of autism spectrum disorder, cerebral palsy, hearing loss, intellectual disability, and vision impairment, metropolitan atlanta, 1991–2010. PloS one, 10(4), e0124120. 10.1371/journal.pone.0124120 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wang G, Divall S, Radovick S, Paige D, Ning Y, Chen Z, Ji Y, Hong X, Walker SO, Caruso D, Pearson C, Wang MC, Zuckerman B, Cheng TL, & Wang X (2014). Preterm birth and random plasma insulin levels at birth and in early childhood. JAMA, 311(6), 587–596. 10.1001/jama.2014.1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wang G, Hu FB, Mistry KB, Zhang C, Ren F, Huo Y, Paige D, Bartell T, Hong X, Caruso D, Ji Z, Chen Z, Ji Y, Pearson C, Ji H, Zuckerman B, Cheng TL, & Wang X (2016). Associationafra Between Maternal Prepregnancy Body Mass Index and Plasma Folate Concentrations With Child Metabolic Health. JAMA pediatrics, 170(8), e160845. 10.1001/jamapediatrics.2016.0845 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Wang X, Zuckerman B, Pearson C, Kaufman G, Chen C, Wang G, Niu T, Wise PH, Bauchner H, & Xu X (2002). Maternal cigarette smoking, metabolic gene polymorphism, and infant birth weight. JAMA, 287(2), 195–202. 10.1001/jama.287.2.195 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.