Prenatal manganese biomarkers and operant test battery performance in Mexican children: Effect modification by child sex

Abstract

Background:

Manganese (Mn) is essential to healthy neurodevelopment, but both Mn deficiency and over-exposure have been linked to prefrontal cortex (PFC) impairments, the brain region that regulates cognitive and neurobehavioral processes responsible for spatial memory, learning, motivation, and time perception. These processes facilitated by attention, inhibitory control, working memory, and cognitive flexibility are often sexually dimorphic and complex, driven by multiple interconnected neurologic and cognitive domains.

Objective:

We investigated the role of child sex as an effect modifier of the association between prenatal Mn exposure and performance in an operant testing battery (OTB) that assessed multiple cognitive and behavioral functional domains.

Methods:

Children (N = 575) aged 6 – 8 years completed five OTB tasks. Blood and urinary Mn measurements were collected from mothers in the 2^nd and 3^rd trimesters. Multiple regression models estimated the association between Mn biomarkers at each trimester with OTB performance while adjusting for socio-demographic covariates. Covariate-adjusted weighted quantile sum (WQS) regression models were used to estimate the association of a Mn multi-media biomarker (MMB) mixture with OTB performance. Interaction terms were used to estimate modification effect by child sex.

Results:

Higher blood Mn exposure was associated with better response rates (more motivation) on the progressive ratio task and higher overall accuracy on the delayed matching-to-sample task. In the WQS models, the MMB mixture was associated with better response rates (more motivation) on the progressive ratio task. Additionally, for the linear and WQS models, we observed a modification effect by child sex in the progressive ratio and delayed matching-to-sample tasks. Higher prenatal Mn biomarker levels were associated with improved task performance for girls and reduced performance in boys.

Conclusion:

Higher prenatal blood Mn concentrations and the MMB mixture predicted improved performance on two of five operant tasks. Higher prenatal Mn concentrations regulated executive functions in children in a sexually dimorphic manner. Higher prenatal Mn exposure is associated with improved performance on spatial memory and motivation tasks in girls, suggesting that Mn’s nutritional role is sexually dimorphic, and should be considered when making dietary and/or environmental intervention recommendations.

Manganese (Mn) is an essential micronutrient critical to the growth and development of the central nervous system and plays a key role in other processes, such as bone development. Normal adult ranges of Mn are 4–15 μg/L in blood and 1–8 μg/L in urine (Martinez-Finley et al., 2013) although levels vary by age and life stage (McRae et al., 2020; Sanders et al., 2015). While Mn is a nutrient needed for neurotransmitter metabolism and regulation of oxidative stress, deficient or excessive Mn exposure can be deleterious to human health, producing maladaptive effects when accumulated in the brain (Haynes et al., 2015). Studies show that prenatal under- and overexposure to Mn in children can adversely interrupt or disrupt normal neurodevelopment, globally impacting motor, psychomotor, learning, and behavioral functions (Chung et al., 2015; de Water et al., 2018; Mora et al., 2015; Yu et al., 2014).

Anatomically, chronic low- and high-level Mn impacts the prefrontal cortex (de Water et al., 2018), the brain region that serves executive function, which may explain the effects of Mn exposure on impairments in executive functioning (Bouchard et al., 2018; Carvalho et al., 2018; Haynes et al., 2018). Such impairment increases the risk of neurodevelopmental delays in problem-solving, attention, inhibition, and language (Irizar et al., 2021; Lin et al., 2013; Mora et al., 2015). Executive functions are higher-order processes that intersect cognition and behavior, including attention, inhibitory control, working memory, motivation, and cognitive flexibility (Diamond, 2013). The role of the effect of Mn in influencing these complex neurobehavioral systems has received little attention in children. There are however, some rodent studies examining how Mn affects executive functioning that utilize various behavioral procedures (Brook et al., 1999; Burkey & Nation, 1994; Morris-Schaffer et al., 2019; Virgolini et al., 2005).

Evidence from children’s studies suggests a sexual dimorphism in the neurotoxic effect of prenatal Mn exposure on neurodevelopment (Chiu et al., 2017; Mora et al., 2015; Takser et al., 2003). However, the data are inconsistent and primarily concentrate on utilizing global survey-based scales (e.g., Child Behavior Checklist, McCarthy Scales of Children’s Abilities, Wechsler Intelligence Scale for Children – Revised, and I.Q. scores; Ericson et al., 2007; Irizar et al., 2021; Rahman et al., 2017; Riojas-Rodríguez et al., 2010) that assesses general cognitive functions relating to intelligence, motor skills, verbal ability, and learning deficits. There has been limited research investigating the role of reward reinforcement using neurobehavioral measures that can directly measure specific cognitive functions. In this study, the National Center for Toxicological Research’s Operant Testing Battery (OTB) was used to assess cognitive and behavioral faculties across various functional domains to assess sexually dimorphic effects. In brief, OTB tasks offer a way to measure behavioral mechanisms directly representing discrete cognitive functions. As a result, they can give us more insight into the overall and specific impacts of neurotoxicants, unlike indirect global survey-based scales.

Therefore, understanding the neurotoxic sexually dimorphic effects of prenatal Mn exposure on executive functions and behavior is critical to inform environmental interventions and cognitive-behavioral developmental programming. We initiated the present study to investigate the role of child sex as an effect modifier on the association between prenatal Mn exposure and childhood neurodevelopment. Thus, we hypothesize that prenatal Mn exposure can impact cognitive and behavioral performance in a sexual dimorphic manner in children.

Methods

Participants

The Programming Research in Obesity, GRowth, Environment and Social Stressors study (PROGRESS) is an ongoing partnership with the National Institute of Public Health and the National Institute of Perinatology in Mexico that has followed 650 children from pregnancy to ages 13–15 years. Although continuing, we analyze children up to 6–8 age, as data are now complete for this visit. Initially, we enrolled 1,084 mothers who delivered a live birth at the same maternity hospital (Hospital de Ginecología y Obstetricia [la Clínica Numero 4]) in Mexico City from 2007 to 2011. Mothers had two prenatal visits, in their 2^nd and 3^rd trimester, with inclusion criteria selecting for healthy >18 years old women who were no more than 20 weeks pregnant, with no cardiovascular or kidney disease, had telephone access, and intended to live in Mexico City for at least three years. Further details on the inclusion criteria have been previously described (Braun et al., 2014). All study protocols were approved by the institutional review boards of the Icahn School of Medicine at Mount Sinai, the Harvard T.H. Chan School of Public Health, the National Institute of Public Health Mexico, and the National Institute of Perinatology Mexico. There are currently 634 actively followed children with data ongoing at ages 13–15. Children (N = 575) included in the study were those who completed the OTB measures at 6–8 years of age.

Blood and Urinary Manganese Measurement

Blood and urine samples were collected from mothers in the 2^nd (between the 16^th and 20^th gestational weeks) and 3^rd trimesters (between the 30^th and 34^th gestational weeks) of pregnancy. Mn for each time point was analyzed with a triple quadrupole dynamic reaction cell-inductively coupled plasma mass spectrometer (ICP-MS; Elan 6100; PerkinElmer, Norwalk, CT) using previously described techniques and quality control procedures (Henn et al., 2010).

Neurodevelopment Assessment of Operant Test Battery

Children ages 6–8 years were administered the OTB, which consists of five behavioral tasks that provoke responses conditional upon distinctive cognitive and neurobehavioral functions (Paule et al., 1999). Further details on the description of the apparatus have been previously provided (Paule et al., 1988). During the tasks, the child was seated across from a testing panel installed in a wooden cabinet (Figure 1). The testing panel had circular press-plates and response levers that the child could press, stimulus lights that could indicate a correct or incorrect response, and a row of 6 lights that could indicate which lever to press for one of the behavioral tasks. Located at the bottom of the apparatus was a container where the apparatus delivered nickels following a correct response. After completing the study visit, children traded the earned nickels for a toy they liked.

Figure 1.

Diagram of child positioning and the operant test panel. Illustration by Jill Gregory used with permission of ©Mount Sinai Health System. Starting at the top of the panel, there is a speaker and below that, there are three circular press-plates. Below the circular press-plates, there are two different types of stimulus lights – correct and incorrect response indicator lights (smiley face) and serial position indicator lights (colored rectangles). At the bottom of the apparatus, there was a container where a dispenser mounted inside the wooden cabinet delivered nickels.

A trained psychologist administered all behavioral tasks, and before each task, children watched video instructions in Spanish. After the children acknowledged they understood the video instructions, the psychologist left the study room and continuously monitored the children’s behavior through a one-way mirror. Children completed the progressive ratio (PR) test that assesses motivation, the conditioned-position responding (CPR) test that assesses color and position discrimination, a temporal response differentiation (TRD) test that assesses time perception, a delayed matching-to-sample (DMTS) task that assesses working memory and attention, and a test of incremental repeated acquisition (IRA) that assesses learning behavior. Table 1 presents the description of each OTB task and five behavioral variables the current study used as primary outcomes.

Table 1.

Description of each Five Behavioral Tasks in the Operant Test Battery

Operant Tasks	Description of Tasks	Outcome Variables
a Progressive-Ratio	The child was required to increase the work performed for each reinforcement. Only the far right retractable level was used in this test. To begin with, small numbers of lever presses (10 lever presses) result in the delivery of nickel. After each nickel was obtained, the response requirement was increased such that subsequent reinforcers were obtained only after a greater number of level presses were made. Thus, the second reinforcer was obtained after 20 level presses, and the third was obtained after 30 lever presses, and so on. The response requirement was continually increased until the task timed out (Paule et al., 1988).	Response rate – the number of responses (lever presses) the child made per minute.
b Conditioned Position Responding	Only the three circular press plates were used. At the beginning of each trial, the center press plate was illuminated with one of four colors – red, yellow, blue, or green. The child pressed the illuminated press-plate and it was darkened, after which the two side press-plates were illuminated white. If the center plate color was red or yellow, then a press on the left-side plate was correct. A press on the right-side plate was correct if the center plate color was blue or green.	Overall observing response latency – the average time children took to respond on the center press-plate after it was illuminated
a Temporal Response Differentiation	The child was required to hold the far-left lever down for 10 to 14 seconds repeatedly for the duration of the session. Lever hold durations were divided into timing holds (holds greater than or equal to 2 s in duration) and bursts (holds less than 2 s in duration; Chelonis et al., 2004).	Standard deviation of the duration of timing hold – the variability of the length of time between the press and release of the level for lever holds that were greater than or equal to two seconds in duration.
b Delayed Matching-to-Sample	At the beginning of each trial, the center press-plate was illuminated with one of seven possible geometrical shapes (Chelonis et al., 2014). The child pressed the center press-plate to extinguish the shape and initiate the delay period (1 to 32 s). After the delay, all three plates were illuminated with different shapes, one of which matched the original shape. The child was required to press the plate that “matched” the geometrical shape they saw previously.	Overall accuracy – the total number of correct responses divided by the number of trials completed across all the delays.
a Incremental Repeated Acquisition	All four levers, the correct and incorrect response indicator, and the serial position indicator lights were used (Baldwin et al., 2012). Briefly, the child was required to learn to press the levers in a specific sequence dictated by the illumination of the serial position indicator lights. The child started with a one-lever sequence and once the child completed 3 correct sequences, the task difficulty was increased to a two-lever sequence. As the child continued to master the task, more lever responses were added to the sequence until the child was performing a six-lever sequence (i.e., a six-link response chain).	Complete chain accuracy – the number of errorless response chains completed divided by the total number of chains completed.

Note. OTB tasks are listed in order administered to children. Type of Manipulanda: ^alever or ^bpress-plate

Statistical Analysis

All statistical analyses were conducted using RStudio 4.0.3 software using the lm function. The interquartile range (IQR) method was used to identify outliers in the dataset, with limits of 1.5 IQR from the first and third quartiles. Data points falling outside these limits were subsequently removed. Observations removed in the following biomarkers included blood and urine Mn at 2^nd (3.3% and 10.7%) and 3^rd (3.7% and 11.4%) trimesters, and in the following OTB tasks, CPR (8.7%) and TRD (7.4%). Descriptive statistics (means, standard deviations (SDs), frequencies, and percentages) were calculated for all variables. Regression models were used to estimate associations between Mn biomarkers at each trimester with OTB performance. Blood Mn was highly correlated across time points, and to avoid multi-collinearity, each was analyzed in separate regression models (See results). In addition, because we integrated blood and urine Mn biomarkers, we used weighted quantile sum (WQS) regression in the multi-media biomarker (MMB) approach described by Levin-Schwartz et al. 2020. This approach integrates biomarkers measured in different media but representing the same chemical. Because the WQS methods use quantiles, measures can be on different scales, as they are common when comparing blood vs. urine biomarkers. For each model, two-way interaction terms were used to estimate the modification effect of child sex. We also applied the false discovery rate (FDR) approach to assess multiple comparisons in all models.

Further, the MMB provides data on the relative contribution of the blood vs. urine biomarker via regression weights without discarding the adding information from having two exposure biomarkers and negating the need to choose between the two biomarkers. A covariate-adjusted WQS regression using the gWQS (Renzetti et al., 2021) package was performed to estimate and examine the cumulative effect of combined Mn biomarkers using a MMB approach proposed by Levin-Schwartz et al. (2020) on OTB measures. All Mn biomarkers measured in blood and urine were included in the mixture. Further details of WQS regression have been previously described (Carrico et al., 2015; Levin-Schwartz et al., 2021). Briefly, the WQS implementation involves two steps: (1) a weighted index representing the association between each Mn biomarker, and the OTB measure will be estimated across 100 bootstrap datasets, and (2) this weighted index will be then tested in a linear regression model estimating the association between the MMB index and the OTB measures.

Models were adjusted for maternal educational attainment (< high school, high school, and > high school), maternal age in the 2^nd trimester, and socioeconomic status (SES) index. Our SES index was calculated based on an index created by the Asociación Mexicana de Agencias de Investigación de Mercados y Opinión Pública (AMAI). We used 13 variables derived from a questionnaire to classify participant families into six levels, which we simplified into a relative three-level index of low, medium, and high SES (Carrasco, 2002). Children’s Mn levels in blood concentration at age 4 were adjusted in each model. Creatinine levels in children’s urinary concentration were used as a covariate to adjust for differences in hydration status at the time of collection in the urine Mn models. A creatinine detection kit (Arbor Assays) evaluated urinary creatinine levels. Maternal blood lead (Pb) levels were adjusted in the blood Mn models, and maternal urinary Pb levels were adjusted in the urine Mn models.

Sensitivity Analyses

Sensitivity analyses were also implemented. First, we performed regression models, including a quadratic term for each Mn biomarker, to investigate nonlinear associations between Mn biomarkers at each trimester and OTB performance. Subsequently, we utilized covariate-adjusted WQS models that intentionally included outliers from each Mn biomarker to provide a more comprehensive representation of the underlying data variability. This approach enhances the reliability of our results. Using quintiles, the WQS method also offers safeguards against extreme values, similar to other procedures like winsorization.

Results

Demographics

Table 2 presents the overall and sex-stratified socio-demographic characteristics of participants, concentrations of Mn biomarkers, and summary statistics for OTB outcome measures. The sample consisted of 575 boys (51%) and girls (49%) between the ages of 6 and 8 years, with a mean age of nearly 7 years. In PROGRESS, maternal education was well distributed across three levels (41% with less than a high school diploma, 36% with a high school diploma, and 23% with more than a high school diploma), with over half (53.6%) being from low SES backgrounds. The mean ± SD for blood Mn in the 2^nd trimester and 3^rd trimester were 14.12 μg/L ± 4.31 and 18.53 μg/L ± 5.38, respectively. In the 2^nd and 3^rd trimesters, the mean for urinary Mn was 1.32 μg/L ± 0.39 and 1.19 μg/L ± 0.40, respectively. Table 2 also presents the average performance on each of the OTB tasks.

Table 2.

Overall and sex-stratified socio-demographic characteristics, Mn biomarkers, and OTB tasks (N = 575)

Children’s characteristics	Overall (N = 575) Mean ± SD or %	Boys (n = 293) Mean ± SD or %	Girls (n = 282) Mean ± SD or %
Age (years)	6.66 ± 0.71	6.76 ± 0.60	6.73 ± 0.60
Maternal Education
< High School	41.0%	43.0%	38.3%
High School	36.0%	33.8%	37.9%
> High School	23.0%	23.2%	23.8%
Maternal SES
Low	53.6%	54.3%	52.8%
Medium	36.3%	35.5%	37.2%
High	10.1%	10.2%	9.9%
Mn Biomarkers
Blood Mn at 2nd trimester (μg/L)	14.12 ± 4.31	14.19 ± 4.30	14.05 ± 4.34
Blood Mn at 3rd trimester (μg/L)	18.53 ± 5.38	18.67 ± 5.48	18.40 ± 5.29
Urine Mn at 2nd trimester (μg/L)	1.32 ± 0.39	1.31 ± 0.38	1.34 ± 0.40
Urine Mn at 3rd trimester (μg/L)	1.19 ± 0.40	1.19 ± 0.40	1.19 ± 0.40
OTB Measures
Response rate (responses/minute)	139.87 ± 46.26	144.46 ± 46.51	135.10 ± 45.60
Overall observance response latencies (seconds)	2.50 ± 0.90	2.39 ± 0.89	2.61 ± 0.90
Timing Hold (SD)	7.94 ± 5.61	7.52 ± 5.37	8.37 ± 5.82
Overall accuracy (percent)	83.99 ± 14.18	83.66 ± 13.73	84.35 ± 14.66
Complete chains accuracy (percent)	40.29 ± 22.82	41.93 ± 22.58	38.57 ± 22.98

Blood Mn and Operant Test Battery Performance

In the response rate model, for 2^nd trimester Mn, there was a significant association between higher blood Mn and higher response rates (β = 1.99, 95% CI: 0.42, 3.56). We observed an association between child sex (β = 65.52, 95% CI: 31.13, 64.92) and response rates per minute, with boys having a 65.5 higher number of responses per minute than girls. There was a significant interaction between blood Mn in the 2^nd trimester and child sex (β = −3.53, 95% CI: −5.67, −1.40). Because boys were coded as a “1,” the interaction means that associations between Mn exposure and response rates were positive in girls and negative in boys (Figure 2a). For Mn levels in the 3^rd trimester, there was also an association between child sex (β = 53.83, 95% CI: 19.67, 88.00) and response rates, indicating that boys made 53.8 higher responses per minute than girls. We observed a significant interaction between blood Mn in the 3^rd trimester and child sex (β = −2.11, 95% CI: −3.88, −0.34), with Mn being positively associated with response rates in girls and negatively associated with response rates in boys (Figure 2b). There was an association between blood Mn and response rates. For this model, we found that a 1-unit increase in blood Mn exposure was associated with a 1.75 higher number of response rates (β = 1.75, 95% CI: 0.42, 3.08).

Figure 2a.

Covariate-adjusted model associating blood Mn levels at the 2^nd trimester and response rates in the Progressive Ratio task

Figure 2b.

Covariate-adjusted model associating blood Mn levels at the 2^nd trimester and overall accuracy in the Delayed Matching-to-Sample task

In the overall accuracy model for Mn levels in the 2^nd trimester, we observed significant associations between blood Mn (β = 0.58, 95% CI: 0.08, 1.08) and child sex (β = 12.73, 95% CI: 2.69, 22.78) with overall accuracy. Overall accuracy increased by 0.58 percent for every 1-unit increase of blood Mn exposure during pregnancy, and boys had a 12.7 higher accuracy percentage than girls. There was a significant interaction between blood Mn in the 2^nd trimester and child sex (β = −0.94, 95% CI: −1.63, −0.26), with Mn being positively associated with accuracy in girls and negatively associated with accuracy in boys (Figure 2c). Further, we observed no statistically significant association with blood Mn measured in the 3^rd trimester on overall accuracy and no effect on child sex and interaction effect. Table 3 shows no significant associations between blood Mn levels at each trimester and the observing response latency, SD of timing hold, and complete chain accuracy models. We also observed no effect on child sex or any significant interaction between Mn exposure during each trimester and the OTB tasks based on the child’s sex. Lastly, in the PR and DMTS models, we evaluated the p-values adjusted for FDR and found that both interaction effects sustained significance in the 2^nd trimester. However, no statistically significant FDR p-values emerged from the remaining models.

Figure 2c.

Covariate-adjusted model associating blood Mn levels at the 3^rd trimester and response rates in the Progressive Ratio task

Table 3.

Association between prenatal blood Mn exposure and Operant Test Battery tasks in 6 to 8-year-olds

Outcome measures	Exposure	Main Effect (Blood Mn)			Main Effect (Child Sex)			Interaction
		B (95% CI)	P Raw	P FDR	B (95% CI)	P Raw	P FDR	B (95% CI)	P Raw	P FDR
PR
Response rate	Blood Mn at 2nd Trimester	1.99 (0.42, 3.56)	0.01	0.03	62.52 (31.13, 93.91)	0.001	0.008	−3.53 (−5.67, −1.40)	0.001	0.008
	Blood Mn at 3rd Trimester	1.75 (0.42, 3.08)	0.01	0.07	53.83 (19.67, 88.00)	0.002	0.03	−2.11 (−3.88, −0.34)	0.02	0.10
CPR
Overall observance response latencies	Blood Mn at 2nd Trimester	−0.00 (−0.03, 0.03)	0.92	0.99	−0.53 (−1.16, 0.09)	0.09	0.19	0.02 (−0.03, 0.06)	0.46	0.77
	Blood Mn at 3rd Trimester	−0.01 (−0.04, 0.01)	0.38	0.55	−0.54 (−1.22, 0.15)	0.13	0.43	0.02 (−0.02, 0.05)	0.40	0.55
TRD
Timing Hold (SD)	Blood Mn at 2nd Trimester	0.00 (−0.19, 0.19)	0.99	0.99	0.83 (−2.98, 4.64)	0.67	0.83	−0.08 (−0.34, 0.17)	0.53	0.80
	Blood Mn at 3rd Trimester	−0.12 (−0.29, 0.05)	0.16	0.43	−2.32 (−6.66, 2.02)	0..29	0.55	0.11 (−0.12, 0.33)	0.36	0.55
DMTS
Overall accuracy	Blood Mn at 2nd Trimester	0.58 (0.08, 1.08)	0.02	0.05	12.73 (2.69, 22.78)	0.01	0.03	−0.94 (−1.63, −0.26)	0.01	0.03
	Blood Mn at 3rd Trimester	0.20 (−0.25, 0.65)	0.38	0.55	2.64 (−8.89, 14.16)	0.65	0.69	−0.16 (−0.76, 0.44)	0.60	0.69
IRA
Complete chains accuracy	Blood Mn at 2nd Trimester	0.39 (−0.36, 1.14)	0.31	0.58	3.66 (−11.46, 18.77)	0.64	0.84	−0.10 (−1.12, 0.93)	0.86	0.99
	Blood Mn at 3rd Trimester	0.45 (−0.19, 1.09)	0.17	0.43	4.73 (−11.72, 21.17)	0.57	0.69	−0.06 (−0.91, 0.79)	0.89	0.89

Note. Child sex (females served as the reference group). Controls are maternal age, maternal education (less than high school served as the reference group), SES (low SES served as the reference group), child urinary creatinine levels, child blood Mn levels, and maternal urinary lead levels. B = Unstandardized Regression Weight; CI = Confidence Interval. p_Raw = raw p-value; p_FDR = adjusted p-value for multiple testing by controlling a false discovery rate (FDR) of .05.

Urine Mn and Operant Test Battery Performance

In the response rate model, we only observed an association between child sex and the number of responses per minute during the 3^rd trimester (β = 30.63, 95% CI: 0.71, 60.54), indicating that boys had 30.6 higher response rates than girls. However, in all other models, no significant associations were observed between urine Mn levels at each trimester and OTB tasks. We likewise observed no significant differences in child sex or interaction by child sex with urine Mn at each trimester on OTB tasks (see Table 4).

Table 4.

Association between prenatal urine Mn exposure and Operant Test Battery tasks in 6 to 8-year-olds

Outcome measures	Exposure	Main Effect (Blood Mn)			Main Effect (Child Sex)			Interaction
		B (95% CI)	P Raw	P FDR	B (95% CI)	P Raw	P FDR	B (95% CI)	P Raw	P FDR
PR
Response rate	Urine Mn at 2nd Trimester	−8.46 (−25.30, 8.38)	0.33	0.53	25.14 (−8.59, 58.88)	0.14	0.53	−10.16 (−34.74, 14.42)	0.42	0.63
	Urine Mn at 3rd Trimester	3.09 (−13.12, 19.29)	0.71	0.89	30.63 (0.71, 60.54)	0.04	0.49	−14.24 (−37.37, 8.89)	0.23	0.49
CPR
Overall observance response latencies	Urine Mn at 2nd Trimester	0.35 (0.00, 0.69)	0.05	0.53	−0.39 (−1.08, 0.30)	0.27	0.53	0.11 (−0.39, 0.61)	0.67	0.81
	Urine Mn at 3rd Trimester	−0.21 (−0.53, 0.11)	0.20	0.49	−0.46 (−1.05, 0.14)	0.14	0.49	0.10 (−0.36, 0.56)	0.67	0.89
TRD
Timing Hold (SD)	Urine Mn at 2nd Trimester	−0.66 (−2.86, 1.55)	0.56	0.72	−2.64 (−7.08, 1.79)	0.24	0.53	1.74 (−1.48, 4.96)	0.29	0.53
	Urine Mn at 3rd Trimester	−0.31 (−2.40, 1.78)	0.77	0.89	−0.59 (−4.48, 3.30)	0.77	0.89	0.29 (−2.74, 3.32)	0.85	0.91
DMTS
Overall accuracy	Urine Mn at 2nd Trimester	0.21 (−5.10, 5.53)	0.94	0.94	−7.13 (−17.78, 3.53)	0.19	0.53	4.58 (−3.18, 12.34)	0.25	0.53
	Urine Mn at 3rd Trimester	1.71 (−3.81, 7.23)	0.54	0.89	6.71 (−3.48, 16.90)	0.20	0.49	−6.72 (−14.60, 1.16)	0.09	0.49
IRA
Complete chains accuracy	Urine Mn at 2nd Trimester	−4.45 (−12.74, 3.83)	0.29	0.53	1.45 (−15.36, 18.27)	0.87	0.94	0.77 (−11.47, 13.01)	0.90	0.94
	Urine Mn at 3rd Trimester	−6.19 (−14.18, 1.79)	0.57	0.89	4.24 (−10.54, 19.02)	0.11	0.49	0.10 (−11.31, 11.50)	0.98	0.98

Note. Child sex (females served as the reference group). Controls are maternal age, maternal education (less than high school served as the reference group), SES (low SES served as the reference group), child urinary creatinine levels, child blood Mn levels, and maternal urinary lead levels. B = Unstandardized Regression Weight; CI = Confidence Interval. p_Raw = raw p-value; p_FDR = adjusted p-value for multiple testing by controlling a false discovery rate (FDR) of .05.

MMB mixture and Operant Test Battery Performance

Our previous study reported on prenatal metal mixtures and PR, including Mn (de Water et al., 2022). Notably, our prior analysis did not use the MMB approach, which integrates exposure over time and across different biomarker matrices (Levin-Schwartz et al., 2020). This analysis builds on that prior work by using the MMB to better estimate Mn effects. In the prior study, the Mn effects were small relative to the lead, zinc, and arsenic effects (de Water et al., 2022). As shown in Table 5, the WQS models identified significant interaction effects with the MMB mixture on the response rate (β = −25.43, 95% CI: −41.05, −9.80) on the PR task and overall accuracy performances (β = −5.86, 95% CI: −10.24, −1.48) on the DMTS task, indicating that the MMB mixture was positively associated with both tasks in girls and negatively associated with both tasks in boys (Figure 3a and Figure 4a). Further, in the response rate model, we observed an association between the MMB mixture (β = 13.81, 95% CI: 2.74, 24.88) and child sex (β = 68.46, 95% CI: 33.18, 103.74) on the number of responses per minute. There was a 13.8 higher response per minute with each quintile increase in the MMB mixture; boys made 68.5 higher response per minute than girls. Maternal blood Mn in the 2^nd trimester and urinary Mn in the 3^rd trimester were the most significant weights to the MMB effect (45.1% and 34.2%; Figure 3b). In the overall accuracy performance model, we observed an association between child sex (β = 10.38, 95% CI: 0.63, 20.13) and overall accuracy, indicating that boys had a 10.38 higher accuracy percentage than girls. Maternal urinary and blood Mn in the 2^nd trimesters were the most significant weights to the MMB effect (46.6% and 34.9%; Figure 4b). In the overall observing response latency model, there was a significant association between overall observing response latency and child sex (β = −0.75, 95% CI: −1.37, −0.13), with boys tending to have a 0.75-second shorter observing response latency than girls. Lastly, we observed no statistically significant association between the MMB mixture and other OTB tasks and no differences in child sex and interaction effects. We assessed the p-values adjusted for FDR in the PR and DMTS models and discovered that the interaction effects remained significant.

Table 5.

Association between Mn MMB mixture and Operant Test Battery tasks in 6 to 8-year-olds

Outcome measures	Exposure	Main Effect (MMB)			Main Effect (Child Sex)			Interaction
		B (95% CI)95% CI	P Raw	P FDR	B (95% CI)	P Raw	P FDR	B (95% CI)	P Raw	P FDR
PR
Response rate	MMB Mixture	13.81 (2.74, 24.88)	0.01	0.04	68.46 (33.18, 103.74)	0.0002	0.003	−25.43 (−41.05, −9.80)	0.002	0.02
CPR
Overall observance response latencies	MMB Mixture	−0.03 (−0.22, 0.16)	0.76	0.76	−0.75 (−1.37, −0.13)	0.02	0.06	0.18 (−0.09, 0.46)	0.20	0.30
TRD
Timing Hold (SD)	MMB Mixture	−0.58 (−1.77, 0.61)	0.34	0.46	−3.03 (−6.92, 0.85)	0.13	0.28	0.69 (−0.92, 2.31)	0.40	0.48
DMTS
Overall accuracy	MMB Mixture	2.23 (−0.83, 5.29)	0.16	0.30	10.38 (0.63, 20.13)	0.03	0.08	−5.86 (−10.24, −1.48)	0.009	0.04
IRA
Complete chains accuracy	MMB Mixture	2.14 (−2.99, 7.27)	0.42	0.48	10.68 (−5.80, 27.16)	0.20	0.30	−2.72 (−9.96, 4.52)	0.46	0.49

Note. Child sex (females served as the reference group). Controls are maternal age, maternal education (less than high school served as the reference group), SES (low SES served as the reference group), child urinary creatinine levels, child blood Mn levels, and maternal blood and urinary lead levels. B = Unstandardized Regression Weight; CI = Confidence Interval. p_Raw = raw p-value; p_FDR = adjusted p-value for multiple testing by controlling a false discovery rate (FDR) of .05.

Figure 3.

a. Covariate-adjusted model examining the moderating effect of child sex on the association between MMB mixture and response rates in the Progressive Ratio task. b. Covariate adjusted estimated weights for each component of the MMB mixture contributing to the integrated effect on response rate performances among children. Higher weights indicate a greater contribution. The dashed line indicates the cut-off value for defining the elements with significant weights in the MMB mixture.

Figure 4.

a. Covariate-adjusted model examining the moderating effect of child sex on the association between MMB mixture and overall accuracy in the Delayed Matching-to-Sample task. b. Covariate adjusted estimated weights for each component of the MMB mixture contributing to the integrated effect on overall accuracy performances among children. Higher weights indicate a greater contribution. The dashed line indicates the cut-off value for defining the elements with significant weights in the MMB mixture.

Sensitivity Analyses Results

The sensitivity analyses yielded no indication of significant nonlinear effects, and results were subsequently not shown. The covariate-adjusted WQS regression models, which considered outliers from each Mn biomarker, were consistent with the primary WQS models. These models showed significant main effects with the MMB index and an interaction effect by child sex for both PR and DMTS tasks, as detailed in Supplemental Table S1.

Discussion

In the current study, we investigated sexually dimorphic effects on the prospective association between prenatal Mn exposure and later-life childhood neurodevelopment using an OTB among children living in Mexico City. In a previous publication, we focused on blood Mn as a component of a larger mixture (de Water et al., 2022). In that paper, we had not yet measured urine Mn. Motivated by recent work on MMBs, we repeated our analysis for Progressive Ratio using the MMB approach and found more robust Mn effects when integrating urine and blood Mn biomarkers. We believe this is further evidence that the strength of MMBs is a more accurate measure of body burden than individual biomarkers of exposure. Our results are also unique in employing operant testing procedures in children in a neurotoxicologically framework. These procedures explicitly involve reward reinforcement, in this case – nickels (Chelonis et al., 2004; Paule et al., 1988), which could be used to purchase prizes. This is distinct from many computerized tasks that do not provide tangible rewards and, therefore, may provide inconsistent or inadequate insight into associative learning processes (Achenbach & Edelbrock, 1991; Robbins & Sahakian, 2002; Wechsler, 2014). Our results indicate that the effects of prenatal Mn exposure are sexually dimorphic, such that increased Mn exposure was associated with improved OTB performance in girls and reduced performance in boys. Second, more generally, we found that prenatal Mn biomarkers were associated with greater motivation/improved performance on two of the five operant tasks measured at age 6 – 8 years, suggesting that prenatal Mn plays a role in the development of brain regions related to these tasks. We observed direct associations between blood Mn in each trimester and a higher number of response rates (PR task) and blood Mn in the 2nd trimester and higher overall accuracy (DMTS task). Third, using an integrated MMB, we also found a sexually dimorphic effect with a decreased performance in boys and an improved performance in girls on the PR and DMTS tasks. Fourth, we observed a pattern of associations between child sex and OTB tasks. Lastly, after conducting sensitivity analyses, there were no significant nonlinear effects. This indicates that the relationship between prenatal Mn and OTB performances remains linear, primarily within the range of Mn levels examined. Additionally, the WQS models, including extreme values, were consistent with the primary models.

Furthermore, in both the PR and DMTS tasks, the effect of prenatal blood and urinary Mn exposure was positive (increased motivation in PR) for girls and negative for boys. Overall, our findings suggest that prenatal Mn concentrations impact the development of cognitive and behavioral functions in children in a sexually dimorphic manner. Higher prenatal Mn exposure is associated with better performance in domains of spatial memory and greater motivation in girls than in boys. This suggests that boys and girls may have distinct biological responses to Mn exposure; Mn’s nutritional role benefits girls, and sex is an essential biologic moderator of these effects. It is important to take note that differences in performances by child sex could be due to greater salience in girls vs. boys as well in the context of Mn exposure.

The sexually dimorphic impact of chemicals has become of great interest in recent years, with NIH now mandating that all applications address sex as a biological variable. In toxicology, it was previously common practice to restrict studies to a single-sex to remove the sexually dimorphic effects of toxic compounds. Today that practice is no longer acceptable, and studies conducted in human populations now commonly assess the role of sex-specific effects. We note that sex-specific effects may be predicated on exposure (i.e., the chemical impacts each sex differently) or may differ based on the phenotype (i.e., the phenotype at baseline differs by sex). For example, spatial working memory is commonly observed to be sexually dimorphic, with boys performing faster than girls in many spatial memory tests. Understanding chemical exposure effects requires understanding the baseline sex-specific effect to avoid misinterpretation. When the phenotype is sexually dimorphic at baseline, which is true for both the PR and DMTS tasks, the interpretation of results should factor in the baseline differences. Only by knowing the baseline difference in the phenotype of interest can one understand how the effect of chemical exposure may differ by sex.

Previous studies observed sex differences among children in OTB performance but did not address the role of environmental factors such as Mn. For example, in the PR task, in a sample of American children, boys performed better than girls (Chelonis et al., 2011), further validating our findings and revealing that children’s cognitive performance may be consistent across different cultures. Even without considering environmental factors, such as Mn, many OTB tasks are sexually dimorphic. Our results provide insights into the underlying mechanisms that drive sexually dimorphic behavior. In that light, we sought to address whether higher Mn in blood reduced the baseline sexual dimorphism or increased it. Our findings confirmed that higher Mn appeared to reduce the overall sexual dimorphism of several cognitive and behavioral functions. Prior work has also determined that prenatal Mn’s neurotoxic effects can adversely impact cognitive and behavioral faculties such as motor functioning (Chiu et al., 2017), short-term memory (Lucchini et al., 2019), executive functioning (Bouchard et al., 2018; Carvalho et al., 2018; Haynes et al., 2018), I.Q. (Zhou et al., 2020), and externalizing and attention behaviors (Ericson et al., 2007; Menezes-Filho et al., 2014). Such assays are global assessments that rely on multiple interconnected cognitive domains, which we were able to evaluate in the current study, specifically through differing OTB tasks.

In addition, growing evidence has indicated a sexually dimorphic effect in the association between prenatal Mn and various general cognitive and behavioral faculties, with stronger effects in girls than boys (Baure et al., 2017; Gunier et al., 2015; Irizar et al., 2021; Roels et al., 2012). However, children’s studies have provided inconsistent positive and/or negative indications of Mn neurotoxicity (Carvalho et al., 2022; Chung et al., 2011; Claus Henn et al., 2017; Irizar et al., 2021; Lin et al., 2013; Mora et al., 2015). Animal studies have also shown that prenatal Mn exposure may occur in a sexually dimorphic manner. Exposure may have different effects on behavioral functions in males in comparison to female mice (Moreno et al., 2009) and may induce a sex-depended remodeling in neuron morphology due to sensitivity differences (Madison et al., 2011). Such inconsistencies may ultimately reveal that the discrete mechanisms behind the sexual dimorphism in Mn neurotoxicity are complex. Perhaps it is due to the sex-specific differences relating to modifications occurring in the neurochemistry (Cosgrove et al., 2007) and overall maturation of the brain (De Bellis et al., 2001; Kaczkurkin et al., 2019), regulation of oxidative stress (Llop et al., 2013), or biological hormonal differences (Ngun et al., 2011). These differences may contribute to sexually dimorphic associations between prenatal Mn exposure and neurodevelopment; therefore, further research is warranted.

Toxicology research on the neurobehavioral effects of Mn has rarely studied the role of reward reinforcement and primarily utilized indirect measures of cognitive and behavioral assessments derived from global survey-based instruments, for example, the Bayley Scales of Infant and Toddler Development (BSITD; Gunier et al., 2015), Child Behavior Checklist (CBCL; Ericson et al., 2007), McCarthy Scales of Children’s Abilities (MSCA; Irizar et al., 2021), and the Wechsler Intelligence Scale for Children – Revised (WISC-R; Riojas-Rodríguez et al., 2010). Since Mn is a nutrient, effects can be either beneficial or toxic and depend both on the life stage at which blood Mn is measured as well as the Mn concentration. In addition, effects may be specific to neurocognitive or behavioral domains, requiring researchers to address a range of tasks to fully capture effects. These indirect assessments can deliver useful indicators of global performance but are problematic in interpreting specific, complex cognitive and behavioral functions altered by neurotoxicant exposures (Slikker et al., 2000). Contrarily, the National Center for Toxicological Research’s OTB consists of behavioral tasks that more directly measure specific cognitive and behavioral functions, which can provide insights into neurotoxicants’ global and discrete effects while differentiating learning and behavioral processes from other cognitive mechanisms (Slikker et al., 2000). The tasks of the OTB provide quantifiable and reliable measures of attention, learning, motivation, short-term memory, timing perception ability, and color/position discrimination (Paule et al., 1988); additionally, in non-human research, they are extensively used to assess the effects of drugs and toxicants on cognitive function (Buffalo et al., 1994; Frederick et al., 1995; Mayorga et al., 2000). Evidence from animal toxicology studies utilizing operant testing and comparative behavioral assessments have shown that neurotoxic exposures to lead (Cohn et al., 1993; Cory-Slechta et al., 1998), ethanol (Clausing et al., 1995), and numerous psychoactive substances (Ferguson et al., 1993; Frederick et al., 1997; Morris et al., 1996; Paule et al., 1984; Schulze et al., 1991) can cause biochemical changes in the brain and alter cognitive and behavioral development.

It is important to put our results in the context of the overall study design. As a longitudinal birth cohort with measures of Mn biomarkers collected in pregnancy and OTB measures collected at 6 – 8 years of age, our observed association suggests that Mn levels in pregnancy may lead to persistent behavior changes that are measurable years later in childhood. Further, because Mn is a vital nutrient, yet excess levels can be toxic, our results depend in part on the nutritional status of the population vs. its environmental Mn exposure levels. With regards to Mn toxicity, numerous occupational studies have documented Mn neurotoxicity effects on executive function as well as memory loss, anxiety, nervousness, aggression, impulsive-compulsive behaviors, emotional lability, and sleep disturbances (Rodier, 1955; Sassine et al., 2002; Wennberg et al., 1991). While we did not measure neuronal dopamine levels, the primary mechanisms of Mn neurotoxicity appear to involve increased oxidative damage to neuronal cells (Aschner & Aschner, 1991; Verity, 1999), with the targeting of dopaminergic systems subserving executive function and attention (Bjorklund et al., 2018; Bouchard et al., 2007; Farias et al., 2010). In vitro, Mn promotes auto-oxidation of dopamine (i.e., the Fenton Reaction), which leads to the creation of reactive dopamine quinones (Jomova et al., 2022). Mn also serves as a substrate for the dopamine transporter (Saputra et al., 2016). Mn impacts synaptic dopamine release and transport, and these effects impact dopaminergic synaptic connections in the cortico-striatal-thalamo-cortical loop, which is essential for normal stimulus-response decisions in executive function (Bari & Robbins, 2013). There are substantial reasons to study the impact of elevated developmental Mn exposure as a driver of dysfunction in cortical and sub-cortical brain structures that affect motivation, spatial memory, and attention. We believe that our results likely reflect the role of Mn in early-life dopamine metabolism on later-life child behavior, given this neurotransmitter’s known role in reward-seeking behavior.

Strengths and Limitations

There are several strengths and limitations to this study. To our knowledge, this is the first study investigating operant conditioning in children in reference to prenatal Mn levels using an OTB. OTB allowed for the evaluation of underlying cognitive and neurobehavioral functions and their sensitivity to neurotoxicants and sex. OTB also differentiates learning and behavioral processes from other domains of cognition (Slikker et al., 2000). Further, OTB adds reward sensitivity to the testing, which many behavioral tasks do not. Understanding the role of reward-seeking may be essential for teasing the impact of executive functions on health outcomes such as obesity, as food may be a reward. Future work will focus on understanding the relationship between performance on the OTB and future measures of obesity. Another strength is the evaluation of child sex as an effect modifier on the association between prenatal Mn exposure and childhood neurodevelopment, as there is limited and inconsistent data on its function. In addition, we measured prenatal Mn concentrations in two media biomarkers from the same source across the same time points. Although there is no standard biomarker for measuring Mn exposure, blood and urine measurement has remained useful to reflect recent Mn exposure (Barahona et al., 2022). The half-life duration of blood Mn is reported to be between 10–42 days compared to urine Mn’s 30 hours (Nelson et al., 1993; Roels et al., 1987). This difference may explain why our results show that the effects on OTB tasks differ between blood and urine Mn biomarkers.

One primary strength of our study is our prospective, longitudinal design. Longitudinal designs like ours provide greater insight into causal relationships than cross-sectional studies and are much more difficult to conduct, as they require years of follow-up. Additionally, an important limitation to note is the generalizability of our study to other populations, given that PROGRESS is a relatively homogeneous Mexican cohort. Another limitation is that there may be potential relationships between Mn exposure and other elements associated with neurodevelopment that we did not investigate. Lastly, our study focused on prenatal Mn levels; we lack the ability to directly measure Mn in the fetus, as the mother must be used as a surrogate of exposure for ethical reasons in all studies of the prenatal environment. Furthermore, future studies should consider the integrated effects of Mn exposure across different media biomarkers and critical time windows of prenatal and postnatal exposure when investigating the association between Mn exposure and childhood neurodevelopment.

Conclusions

Our findings show a sexually dimorphic association between prenatal Mn neurotoxic effects on childhood neurodevelopment. More notably, our work extends the literature to include operant tests, commonly used in animal toxicology studies and sparse in human studies. We propose that dopamine metabolism may play a role in these results. Given that identical OTB tasks can be conducted in animals, we propose that future animal research should consider utilizing the OTB metric. Such work would include varying Mn levels and assessing dopamine metabolism in the brain to better examine the mechanisms underlying global and sexually dimorphic cognitive and behavioral functions that arise from prenatal exposure to deficient, sufficient, or excess levels of Mn.

Highlights.

• The effects of prenatal Mn exposure on OTB performance are sexually dimorphic

• Girls had improved spatial memory and motivation tasks when exposed to higher Mn

• MMB mixture on two of five OTB measures differ in the directionality by child sex