Grantee Research Project Results

Final Report: Biomarkers and Neurobehavioral Effects of Perinatal Exposure to Chlorpyrifos and Other Organophosphate Insecticides

EPA Grant Number: R828611
Title: Biomarkers and Neurobehavioral Effects of Perinatal Exposure to Chlorpyrifos and Other Organophosphate Insecticides
Investigators: Wilkins, John R. , Moeschberger, Melvin L. , Weghorst, Christopher M. , Dietrich, Kim , Nishioka, M.
Institution: The Ohio State University , Battelle Memorial Institute , University of Cincinnati
EPA Project Officer: Hahn, Intaek
Project Period: February 12, 2001 through February 11, 2004 (Extended to February 11, 2006)
Project Amount: $1,126,463
RFA: Biomarkers for the Assessment of Exposure and Toxicity in Children (2000) RFA Text |  Recipients Lists
Research Category: Children's Health , Human Health

Objective:

This population-based research project was designed to evaluate the putative relationship between adverse neurobehavioral effects among infants and young children and perinatal exposure to chlorpyrifos (CP), diazinon (DZ), and other organophosphate (OP) insecticides. With the U.S. Environmental Protection Agency’s (EPA) permission, we added to the study protocol an arm that was designed to focus on the potentially harmful neurodevelopmental effects of perinatal exposure to pyrethroid (PYRE) insecticides (full justification available on request). Based on sample size calculations made prior to the submission of our application, it was determined that 176 women early in their second trimester of a low-risk pregnancy should be recruited into the longitudinal study. The 176 figure took into account an estimate of the number of drop-outs anticipated. After birth, the study protocol called for followup of the newborns until they were 2 years of age, with neurobehavioral testing conducted at 3 and 24 months postnatal. In the original application, the stated objectives of this project were:

1. Using appropriate biomarkers of exposure, conduct longitudinal assessments of pre- and postnatal exposures to CP, DZ, and other OP insecticides.
2. Using appropriate biomarkers, perform assessments of pre- and postnatal exposures to other neurodevelopmental toxicants.
3. Using well-standardized/validated instruments, evaluate the neuropsychological competence of the infants and young children by administration of appropriate test batteries.
4. Using well-standardized/validated instruments such as the Parenting Stress Index and the Home Screening Questionnaire, obtain control data on a set of covariates likely to act as confounders and/or effect modifiers.
5. Estimate the genetic susceptibility of the mothers and their children to OP toxicity by determining the paraoxonase genotype of each.

Summary/Accomplishments (Outputs/Outcomes):

Overview of Data Collection

For each pregnancy, attempts were made to obtain the following data: (1) maternal exposure to the OP and pyrethroid insecticides of interest; (2) maternal exposure to other neurodevelopmental toxicants likely to be factors in the target population (e.g., Pb and nicotine); (3) maternal demographics and other potentially confounding family-based factors (e.g., socioeconomic status [SES]); and (4) relevant clinical information pertaining to the pregnancy and birth event (e.g., Apgar scores). Maternal exposures to the OPs and pyrethroids of interest were assessed by analyses of two urine samples obtained from the mother prior to birth. At regular intervals throughout the postnatal follow-up period, urine samples were collected from the infants and analyzed for the OP and pyrethroid metabolites of interest (as will be described). In addition, attempts were made to collect two postnatal blood samples from the youth (one at 12 months, one at 24 months) in order to determine blood lead (Pb) levels. Blood was also used to determine the infant’s paraoxonase (PON1) genotype, a biomarker of susceptibility to OP toxicity. Because vulnerability to the adverse effects of neurodevelopmental toxicants begins shortly after conception, blood was obtained from the expectant mothers to determine the mother’s PON1 genotype.

After birth, relevant data were obtained from the mother-child dyads at 3, 12, and 24 months. At 3 months, neurobehavioral/neurodevelopmental data were obtained on the infant by administration of the Bayley Scales of Infant Development-II (BSID-II) assessment. At 12 months, control data on potential confounders were obtained, in addition to data on breastfeeding and parental IQ. At 24 months, the primary neurodevelopmental data were obtained by a second administration of the BSID-II, in addition to a one-time administration of Ireton’s Child Development Inventory (CDI). Multiple regression modeling is planned to evaluate the relationship between the indicators of neurobehavioral development obtained and OP and PYRE exposure.

Recruitment

As reported previously, recruiting pregnant women into this study proved to be extraordinarily challenging, certainly much more challenging than initially anticipated.

As indicated in the original proposal, we identified three sources of potential participants: (1) the Ohio State University (OSU) Prenatal Clinic; (2) OSU faculty private practice (Stoneridge Women’s Health Center); and (3) the Columbus Neighborhood Health Centers. These three sources were targeted primarily because successful recruiting from all three would ensure a sociodemographically diverse sample, which was what we wanted/needed. All three sources proved to be disappointments. Although we received strong letters of support, persons having authority over management and operation of the OSU Prenatal Clinic proved to be less cooperative than we were led to believe. Some of the barriers encountered here may be related to the timing of our recruitment efforts: Health Insurance Portability and Accountability Act (HIPPA) requirements were in the process of being implemented for the first time then. In addition, many of the participants that we were able to recruit from this source either dropped out or were lost to followup. We had a similar experience at the Stoneridge Women’s Health Center, although we were eventually allowed to engage in “passive recruiting” at that site (inclusion of our study flyer in the packet of materials new obstetric patients receive; not surprisingly, the response rate to this approach to recruiting was very poor). Recruitment at the Columbus Neighborhood Health Centers also proved ineffective, primarily because almost all of these women dropped out or were lost to followup. Upon realization in 2002 that we needed alternative sources of participants, we began posting and/or distributing information about the study in the following places/locations in central Ohio: St. Anne’s Hospital, Mt. Carmel Hospitals, Ohio Health, The Columbus Health Department, Columbus Children’s Hospital (and satellite clinics), daycare centers, toy stores, maternity shops, grocery stores with pharmacies, pharmacies, maternity shops and other childbirth-related organizations, the OSU Student Health Center, private physicians’ offices, the OSU Student Unions, the YWCA, malls, churches, and OSU Family Housing.

As indicated in the various Annual Reports we have submitted, the most efficient approach to recruiting requires private-practice obstetricians willing to “actively” recruit their patients into the study. This amounts to a brief endorsement of the study on the part of the physician and provision of our flyer prior to any contact with study staff. Telephone coordination with each office permits study staff to meet with interested patients right after a late first or early second trimester office visit.

Recruitment Results

The time course of recruitment was graphically visualized, which showed little effect of different recruitment strategies on the recruitment rate. It should be realized that recruiting did not begin until the 15th month of the project period due to late startup and Institutional Review Board (IRB)-related delays in Year 1. Specifically, the OSU Research Foundation was not able to create an account for this project until June 2001, approximately 3 months after the official start date of February 6, 2001. Further, the OSU-based IRB process took about 4 months to complete. After notification of EPA that our protocol had been approved by the OSU IRB, EPA then asked us to investigate Ohio law pertaining to mandated reporting, which required University attorneys to conduct a lengthy search of the relevant Ohio statutes. Because the findings were ambiguous, EPA requested we change our already approved consent form, which required another round of IRB review, which pushed the startup of the project into Year 2.

At the time recruitment was terminated (January 2006), 174 pregnant women had agreed to participate, that is, 174 pregnant women signed consent forms. However, among these 174 recruits, 160 (92.0%) actually entered the study and began participating in the prenatal phase of data collection. By the end of these 160 pregnancies, 2 urine samples had been collected from each of 138 women, while 1 urine sample was collected from each of 22 women, yielding a total of 298 prenatal urine samples. Among the 160 women referred to, 34 dropped out, resulting in a total of 126 healthy singleton births entered into the longitudinal study. Among the 126 singleton births, a total of 351 postnatal urine samples were collected, as follows: 5 samples were collected from 3 infants, 4 samples were collected from 48 infants, 3 samples were collected from 25 infants, 2 from 19, and 1 sample from 31 infants.

Exposure Assessment: Maternal Excretion of Selected OP and PYRE Metabolites During Pregnancy

Table 1 summarizes the metabolites of interest. Note that the acronyms given in Table 1 will be used throughout the rest of this document, and that the measured concentrations of each metabolite in the maternal urine samples (in ng/ml of urine) were converted to excretion measures in ng/day by using urine volume and the duration of the overnight urine collection period. As indicated in the table’s footnotes, DMCA1 is probably cis-DMCA and DMCA2 is probably trans-DMCA (inferred from the fact cis-DCCA elutes before trans-DCCA). This uncertainty stems from the absence of both the cis- and trans- standards for DMCA.

Table 1. OP and PYRE Metabolites of Interest

Metabolite Class	Acronym	Parent Pesticide
1. Phenoxybenzoic PYRE metabolites
3-Phenoxybenzoic acid	3PBA	Cypermethrin, lambda cyhalothrin, deltamethrin, esfenvalerate, permethrin, sumithrin, tralomethrin
4-Fluoro-3-phenoxybenzoic acid	4F-3PBA	Cyfluthrin
2. Divinylcyclopropyl Carboxylic Acid PYRE metabolites
Cis- and trans-3-(2,2-dimethylvinyl)-2,2-dimethylcycplopropane-1-carboxylic acid	DMCA1a DMCA2a	Pyrethrin I (most abundant component of pyrethrin mix), tetramethrin, resmethrin, sumithrin
Cis-and trans-3-(2,2-dichlorovinyl)-2,2-dimethylcycplopropane-1-carboxylic acid	cis-DCCA trans-DCCA	Permethrin, cypermethrin, cyfluthrin
Cis-and trans-3-(2,2-dibromovinyl)-2,2-dimethylcycplopropane-1-carboxylic acid	DBCAb	Deltamethrin
4-chlorophenyl-2-isopropylacetic acid	CIAA	Esfenvalerate
3. Phenolic OP metabolites
3,5,6-trichloro-2-pyridinol	TCPy	Chlorpyrifos
2-isopropyl-6-methyl-4-pyrimidinol	IMPy	Diazinon
a DMCA1 is probably cis-DMCA; DMCA2 is probably trans-DMCA. b The analytical method used cannot distinguish the two DBCA isomers.

Detection rates varied considerably by metabolite. DBCA was not detected in any of the prenatal samples, and DMCA1, DMCA2, and CIAA were rarely detected (< 10% of the samples). On the other hand, TCPy and 3PBA were frequently found, in 96.3% and 92.6% of the samples, respectively. DCCA was detected roughly 50–60% of the time. Approximately 15% of the samples contained measurable levels of 4F-3PBA and IMPy.

Descriptive statistics were estimated/calculated for maternal excretion of the phenolic OP metabolites, for the phenoxybenzoic PYRE metabolites, and for the divinylcyclopropyl carboxylic acid PYRE metabolites, for both trimesters of pregnancy. In addition, cumulative frequency graphs were constructed. A representative example is given below (TCPy and IMPy, both trimesters combined (Figure 1).

Figure 1. Maternal Excretion of TCPy and IMPy During Pregnancy

Exposure Assessment: Infant Excretion of Selected OP and PYRE Metabolites After Birth

As for the maternal data, descriptive statistics were estimated/calculated for infant excretion of the metabolites of interest (Figure 2). Note that here it was possible to estimate measures of excretion on a body weight basis, that is, as ng/kg/day. In addition, data were stratified by the postnatal timing of urine collection, which was planned to occur at 2, 9, 16, and 23 months postnatal. A noteworthy time trend was observed here: the 2-month samples tended to have the lowest concentration values, with higher levels seen at 9 months, and still higher levels at 16 and 23 months, although the 16- and 23-month levels tended to be similar, suggesting a leveling off effect after about 1 year of age. Although not discussed in detail in this report, some preliminary analyses indicate, in addition to the increasing trend just referred to, the presence of a seasonality effect for some analytes, likely explained by the use patterns of those particular compounds.

Figure 2. Urinary Excretion of TCPy (ng/kg/day), by Age of Infant

Exploratory Data Analyses: Urinary Metabolite Levels and 3-Month BSID-II Scores

We first estimated metabolite-specific Pearson Product-Moment correlation coefficients for the 3-month motor and mental BSID-II scores. We did not compute correlation coefficients for those metabolites having more than 80% of their values below the limit of detection (LOD). None of the estimated p-values associated with the relatively small correlation coefficients were found to be < 0.05, although eight were in the expected direction (i.e., eight correlation coefficients have negative values).

Graphs showing the relation between both types of BSID-II scores and the maternal and infant metabolite levels were constructed. Here, box plots of both the average maternal and infant metabolite levels were compared to the BSID-II scores. Again, we did not consider any metabolites with more than 80% of the observations below the LOD. Visual inspection of the plots suggested little to no relation between metabolite levels and infant performance on the 3-month BSID-II (for both mental and motor subscales).

Next, each of the metabolite levels was grouped as follows: 0–20%, 20–40%, 40–60%, 60–80%, 80–90%, 90–100%. Note that this roughly corresponds to dividing the metabolite levels into quintiles; however, due to some of the extreme values of the metabolites, we further divided the highest quintile into two groups in order to more accurately capture potential differences in response at higher levels of metabolites. We then estimated the slope (with 95% confidence interval) for each of the categorized metabolite levels versus mental development index, motor development index, and total (mental + motor) development index scores. We did not include any metabolites with over 80% of the observations below the LOD. Inspection of the associated p-values indicated that statistical significance was rarely reached. However, for both OP metabolites, for both mothers and infants, the direction of the association was generally in the expected direction.

Exploratory Data Analyses: Urinary Metabolite Levels and 24-Month BSID-II Scores

Here, we estimated metabolite-specific Pearson Product-Moment correlation coefficients for the 24-month motor and mental BSID-II scores. Here, we did not compute correlation coefficients for those metabolites having more than 50% of their values below the LOD. Only one of the p-values associated with the correlation coefficients was found to be < 0.05, although 16/28 (57.1%) were in the expected direction (i.e., 16 correlation coefficients had negative values). Coefficients ranged in value from -0.28 to +0.28.

When correlation coefficients for maternal levels of selected metabolites and the 24-month BSID-II scores were estimated, we found strong and statistically significant correlations between the third trimester levels of cis-DCCA, trans-DCCA, and 3PBA and the 24-month mental/cognitive BSID-II index. With the exception of TCPy, all third trimester coefficients were in the expected direction. Values ranged from -0.51 to +0.24.

Graphs showing the relation between both types of BSID-II scores and the maternal and infant metabolite levels were also constructed. Here, box plots of both the average maternal and infant metabolite levels were compared to the BSID-II scores. Note that we did not consider any metabolites with more than 80% of the observations below the LOD.

For both the mothers’ second and third trimester urine samples and the infant urines collected at 2, 9, 16, and 23 months postnatal, we standardized each of the metabolites that had at least 50% of the observations above the LOD. Then, we averaged the z-scores for the metabolites for each mother or infant to obtain one overall average z-score measure. This measure serves as an index measure of overall metabolite levels (a surrogate for overall “exposure”) for each individual. We then estimated the slope across metabolite quintiles to determine if there was a linear trend in either motor or mental MDI scores as metabolite levels increase. Here, statistical significance was reached only once, but most of the slopes were negative, which is in the expected direction.

Scatterplots for the 24-month BSID-II scores and maternal levels of selected metabolites, by trimester of urine sampling, were constructed. As can be seen below (Figure 3), second trimester levels of TCPy appear associated with a notable decline in the 24-month mental and motor scores, while the third trimester levels are not (this graph not shown). For the three PYRE metabolites having a detection rate > 50%, a reversal in this trend is evident. For 3PBA (graph for third trimester shown below; Figure 4), third trimester levels are associated with notable declines in both the mental and motor scores, while the second trimester trendlines are essentially flat (graph not shown). For both isomers of DCCA, second trimester levels were essentially flat, but for the third trimester, a notable decline was seen in the mental score but not the motor scores. These plots serve as a starting point for more in-depth analyses to be conducted in the near future. One question these finding raise is the impact of exposure timing; that is, is there a difference with respect to OPs vis-à-vis the PYREs, and from a toxicological point of view, are the PYRE metabolites likely to influence mental scores but not motor scores?

Figure 3. Second Trimester Maternal Excretion of TCPy and 24-month BSID-II Scores

Figure 4. Third Trimester Maternal Excretion of 3PBA and 24-month BSID-II Scores

Contributions of Project to Advancement of Knowledge

Based on the state-of-the-science at the time, this project was initially designed to address the question of perinatal exposure to OP insecticides and the effect of such exposure on neurodevelopment. Approximately halfway through the project period, EPA agreed that adding a study arm to address the effect of perinatal exposure to pyrethroid (PYRE) insecticides was justified, so we moved in that direction. Overall then, this project addressed the potentially adverse neurodevelopmental effects of perinatal exposure to selected OP insecticides (chlorpyrifos, diazinon) and selected PYRE insecticides (several phenoxybenzoic PYRE insecticides, and several divinylcyclopropyl carboxylic acid PYRE insecticides).

Second, that part of the project that was concerned primarily with exposure assessment produced several noteworthy findings. With respect to prenatal maternal exposures to the compounds of interest, overnight urine sampling was conducted at two points in time: as early in the pregnancy as possible (usually at the beginning of the second trimester), and as late as possible into the pregnancy but before birth. This allowed, for example, exploration of time trends during pregnancy and investigation into the factors that might explain differences between the early and late results. In addition, these data permit consideration of the timing of exposure with respect to the primary neurobehavioral outcome(s) of interest, as discussed below.

One important finding in the context of exposure assessment was that some metabolites were detected in the urine of pregnant women frequently (TCPy and 3PBA were detected in > 90% of the samples), some were detected about half of the time (trans- and cis-DCCA were detected 50-60% of the time), some were detected about 20% of the time (4F-3PBA and IMPy), and some (DMCA and CIAA were rarely found (< 10% of the time); DBCA was not detected in any of the samples). Obviously, when metabolites are not frequently detected, this impairs the ability to assess the health effects of such exposures. Because their detection rates were > 50%, we focused (in the health effects assessment component of the project) on the following urinary analytes: TCPy (detection rate was 96.3%), 3PBA (92.6%), trans-DCCA (58.1%), and cis-DCCA (50.3%)

With respect to exposure assessment efforts that focused on the infants, we implemented a unique urine sampling/collection program. Starting at 2 months of age, mothers mailed in urine-soaked overnight diapers, which were then analyzed by Battelle for the metabolites of interest. After the 2-month mailing, mothers were asked to mail in urine-soaked overnight diapers 3 more times; at 9, 16, and 23 months. Detection rates among the infant urines were similar to those among the mothers. In addition, a noteworthy time trend was observed: the 2-month samples tended to have the lowest concentration values, with higher levels seen at 9 months, and still higher levels at 16 and 23 months, although the 16- and 23-month levels tended to be similar, suggesting a leveling off effect after about 1 year of age. Although not discussed in detail in this report, some preliminary analyses indicate, in addition to the increasing trend just referred to, the presence of a seasonality effect for some analytes, likely explained by the use patterns of those particular compounds.

Neurobehavioral outcomes studied so far include the 3- and 24-month mental and motor BSID-II scores. Not surprisingly, no associations were found between maternal metabolite levels and the 3-month BSID-II scores, or with the 2-month infant urines. However, scatterplots for the 24-month BSID-II scores and maternal levels of selected metabolites, by trimester of urine collection, produced results clearly needing further study/investigation. For example, higher second trimester maternal levels of TCPy appear associated with lower 24-month mental and motor scores, while the third trimester levels were not. For the three PYRE metabolites having detection rates > 50%, a reversal in this trend was evident. For 3PBA, higher third trimester levels were associated with declines in both the mental and motor scores, while the second trimester trendlines were essentially flat. For both isomers of DCCA, second trimester levels were essentially flat, but for the third trimester, a decline was seen for the mental scores but not for the motor scores. These plots serve as a starting point for more in-depth analyses to be conducted in the near future. One question these finding raise pertains to the impact of exposure timing, that is, is there a difference with respect to OPs vis-a-vis the PYREs, and, in addition, from a toxicological point of view, are the PYRE metabolites likely to influence mental scores but not motor scores?

Journal Articles:

No journal articles submitted with this report: View all 2 publications for this project

Supplemental Keywords:

infants, children; insecticides; biomarkers of exposure, susceptibility. , Scientific Discipline, Health, RFA, Susceptibility/Sensitive Population/Genetic Susceptibility, Molecular Biology/Genetics, Toxicology, genetic susceptability, Health Risk Assessment, Children's Health, Biochemistry, Genetics, biomarkers, insecticides, environmental hazard exposures, developmental disorders, health effects, organophosphate pesticides, assessment of exposure, perinatanl exposure, endocrine disruptors, genetic susceptibility, toxics, longitudinal study, infants, sensitive populations, measuring childhood exposure, biological markers, children, adolescence, exposure, vulnerability, chlorpyrifos, children's vulnerablity, neurobehavioral effects, health risks,, RFA, Health, Scientific Discipline, Susceptibility/Sensitive Population/Genetic Susceptibility, Health Risk Assessment, genetic susceptability, Molecular Biology/Genetics, Children's Health, Biochemistry, Toxicology, Genetics, health effects, infants, perinatanl exposure, environmental hazard exposures, health risks, sensitive populations, biological markers, measuring childhood exposure, organophosphate pesticides, chlorpyrifos, genetic susceptibility, insecticides, neurotoxicity, longitudinal study, neurobehavioral effects, children, exposure, children's vulnerablity, endocrine disruptors, toxics, assessment of exposure, adolescence, developmental disorders, vulnerability

Relevant Websites:

Synthesis Report of Research from EPA’s Science to Achieve Results (STAR) Grant Program: Feasibility of Estimating Pesticide Exposure and Dose in Children Using Biological Measurements (PDF) (42 pp, 3.87 MB)

Progress and Final Reports:

Original Abstract

2001 Progress Report

2002 Progress Report

2003 Progress Report

2004 Progress Report