Organophosphate Insecticide Metabolites in Prenatal and Childhood Urine Samples and Intelligence Scores at 6 Years of Age: Results from the Mother–Child PELAGIE Cohort (France)

Background: Several studies suggest that exposure to organophosphate insecticides (OP) during pregnancy impairs neurodevelopment in children. Objectives: We evaluated associations between biomarkers of prenatal and postnatal OP exposure and cognitive function of 6-year-olds in a French longitudinal birth cohort. Methods: In 2002–2006, the PELAGIE mother–child cohort enrolled pregnant women from Brittany. For a random subcohort, we measured nonspecific dialkylphosphate metabolites (DAP) of OP in one maternal urine sample, collected before 19 weeks’ gestation, and in one urine sample collected from their 6-year-old children. Six subtests of the Wechsler Intelligence Scale for Children, 4th edition (WISC-IV) were administered when the children were 6 years of age to evaluate cognitive function (n = 231). Linear regression models controlling for factors including maternal intelligence and the Home Observation for Measurement of the Environment score were used. Results: WISC-IV scores were not significantly associated with prenatal or childhood total DAP metabolites. WISC verbal comprehension score was significantly higher in association with the highest maternal urinary concentrations of diethylphosphate (DE) metabolites (5.5; 95% CI: 0.8, 10.3 for > 13.2 nmol/L vs. < LOQ), whereas WISC working memory score was significantly lower in association with the highest urinary concentrations of DE metabolites at age 6 years (–3.6; 95% CI: –7.8, –0.6 for > 11.1 nmol/L vs. < LOD). Conclusion: We found no evidence that prenatal OP exposure adversely affected cognitive function in 6-year-olds, perhaps because of the population’s socioeconomic status, which was higher than in previous studies, though other causal and noncausal explanations are also possible. The negative association between WISC score and concurrent DE urinary concentrations requires replication by longitudinal studies investigating childhood OP exposure. Citation: Cartier C, Warembourg C, Le Maner-Idrissi G, Lacroix A, Rouget F, Monfort C, Limon G, Durand G, Saint-Amour D, Cordier S, Chevrier C. 2016. Organophosphate insecticide metabolites in prenatal and childhood urine samples and intelligence scores at 6 years of age: results from the mother–child PELAGIE cohort (France). Environ Health Perspect 124:674–680; http://dx.doi.org/10.1289/ehp.1409472


For the maternal samples (by the LABOCEA Institute, in 2008)
Table: Summary of the analytical method characteristics at three levels (for maternal samples).

For the children's samples (by the LABOCEA Institute, in 2013)
Table: Summary of the analytical method characteristics at three levels (for the children's samples). Table S1. Descriptive characteristics assessed at inclusion in the cohort of the families participating (n=231) and not participating (n=246) in the 6-year neuropsychological follow-up (PELAGIE cohort, France) Table S2. Univariate analyses for studying the association between WISC scores of 6-year-old children and covariates (n=231; PELAGIE cohort, France) Table S3. Associations between organophosphate urinary metabolites and WISC working memory scores (n=231; PELAGIE cohort, France) with minimal adjustment Table S4. Associations between organophosphate urinary metabolites and WISC verbal comprehension scores (n=231; PELAGIE cohort, France) with minimal adjustment Table S5. Associations between organophosphate urinary metabolites and WISC working memory scores among participants with complete data (n=216; PELAGIE cohort, France) Table S6. Associations between organophoshate urinary metabolites and WISC verbal comprehension scores among participants with complete data (n=216; PELAGIE cohort, France) Table S7. Prenatal urinary organophosphate metabolite concentrations (in nmol/L) across cohorts addressing the potential role of exposure to organophosphate insecticides during pregnancy on neurodevelopment Figure S1. Spline regressions for studying the associations between organophosphate urinary metabolites and WISC working memory scores (n=231; PELAGIE cohort, France) Figure S2. Spline regressions for studying the associations between organophosphate urinary metabolites and WISC verbal comprehension scores (n=231 ; PELAGIE cohort, France)

For the maternal samples (by the LABOCEA Institute, in 2008)
For this project, we developed a fully automated method for the determination of urinary metabolites of several pesticides simultaneously (triazines, organosphosphates, carbamates, and chloroacetanilides): automated online sample clean-up by solid-phase extraction/liquid chromatography-electrospray ionization tandem mass spectrometry detection offers cleaner and faster sample preparation and analysis, without either matrix signal suppression or peak broadening.

Reagents and chemicals
Reference standards were purchased from Dr. Ehrenstörfer and from Promochem. LC-MS-grade acetonitrile and methanol were purchased from Fisher (>99%). Analytical grade formic acid was bought from Baker (98%). Nitrogen and argon were purchased from Air Liquide at a minimum purity >99%.

Preparation of standard solutions
The standards solutions of each DAP were prepared in methanol. The internal standards were Di-nbutylphosphate and diuron D6 prepared in methanol. All standards and stock solutions were stored at -20 °C until use. Calibration standards were prepared by adding appropriate working standard solutions to 10 mL fresh sample of DAP-free human urine before extraction to obtain concentrations in the range of calibration.

Sample preparation and extraction procedure
After the urine samples were thawed and shaken, the supernatant was analyzed. The samples (5 mL) were preconcentrated by an automated sample preparation system for high sample volume. The online SPE high volume Symbiosis System (Spark Holland, Netherland) is composed of two units: an automatic cartridge exchange (ACE) module, which hold two trays with up to 96 cartridges, and a high pressure dispenser (HPD) module. The ACE unit is equipped with two clamps and two high-pressure valves. While one cartridge is eluting on the right clamp, the next one is being preconditioned in the left clamp.
The Hysphere C18 HD (2×10 mm) was chosen because it yields the best recovery and retention and the most satisfactory peak shape for the largest number of pesticides. The analytes trapped in the cartridges were eluted with the chromatographic mobile phase.

Chromatographic conditions
Chromatographic separation was performed with a reversed phase Synergi fusion-RP analytical column (250 mm × 2.0 mm, 4 µm particle diameter). The mobile phase was a gradient of a mixture of 5 mM ammonium formiate-0.01% formic acid and acetonitrile 0.01% formic acid. The flow rate was 0.2 mL/min. The chromatographic analysis was performed at 35°C.

Mass spectrometry
LC-MS-MS analyses were performed with a system comprising a Waters alliance 2690 LC pump equipped with an autosampler and connected in series with a Quattro Ultima triple-quadrupole mass spectrometer from Micromass ® , UK. The mass spectrometer was equipped with electropray ionization (ESI). Acquisition was performed in the multiple reaction monitoring (MRM) mode, monitoring two transitions per compound (one for quantification and one for confirmation) in positive ionization mode.

Validation study
All validation procedures were performed with fresh samples of triazine-free human urine. The limit of detection (LOD) was defined as the lowest concentration that the analytical process can reliably differentiate from background levels; it was obtained when the signal was three times the background noise in the chromatograph at the lowest analyte concentration assayed. For the limit of quantification (LOQ), the signal must be ten times the background noise. Based on the characterized ion ratio (one quantitative and one confirmation) of each compound, intra-assay precision and accuracy were assessed at 3 levels in the range LOQ-10 µg/L. If the sample was out this range, it was diluted to be in the range of calibration. In all, five replicate quality control samples of each of the three levels of concentrations were analyzed.
The calibration curves showed good linearity with a correlation coefficient >0.990. The method is precise (CV%≤20%) and accurate. Analytical characteristics resulting from the validation of the method are reported in the table below.

For the children's samples (by the LABOCEA Institute, in 2013)
This project used a fully automated method for simultaneous determination of urinary metabolites of pesticides: automated online sample clean-up by solid-phase extraction/Ultra performance liquid chromatography-electrospray ionization tandem mass spectrometry.

Reagents and chemicals
Reference standards were purchased from Dr. Ehrenstörfer and from Cerilliant. LC-MS grade acetonitrile and methanol were purchased from Fisher (>99%). Analytical grade formic acid was bought from Fisher (99%). Nitrogen and argon were purchased from Air Liquide at a minimum purity >99%.

Preparation of standard solutions
The standard solutions of each DAP were prepared in methanol. The internal standards, Diehylpthiophosphate D10 and dimethylthiophosphate D6, were prepared in methanol. All standards and stock solutions were stored at -20 °C until use. Calibration standards were prepared by adding appropriate working standard solutions to 10 mL fresh sample of DAP-free human urine before extraction to obtain concentrations in the range of calibration.

Sample preparation and extraction procedure
After the urine samples were thawed, shaken, and centrifuged, the supernatant was analyzed. The samples (1 mL) were preconcentrated by an automated sample preparation system. The online SPE is a Waters 2777C sample manager. Waters Oasis HLB Direct Connect cartridge (2.1 × 30 mm) was chosen because it yields the best recovery and retention as well as satisfactory peak shape for these metabolites. The analytes trapped in the cartridges were eluted with the chromatographic mobile phase.

Chromatographic conditions
Chromatographic separation was performed with a reversed phase Waters BEH C18 analytical column (150 mm × 2.1 mm, 1.7 µm particle diameter). The mobile phase was a gradient of a mixture of 0.05% formic acid and acetonitrile 0.05% formic acid. The flow rate was 0.3 mL/min. The chromatographic analysis was performed at 40°C.

Mass spectrometry
LC-MS-MS analyses were performed with a system comprising a Waters Acquity UPLC Binary and a Quaternary pump, connected in series with a Xevo TQ-S triple-quadrupole mass spectrometer from Waters. The mass spectrometer was equipped with electropray ionization (ESI). Acquisition was performed in the multiple reaction monitoring (MRM) mode, monitoring two transitions per compound (one for quantification and one for confirmation) in positive ionization mode.

Validation study
All validation procedures were performed with fresh samples of triazine free human urine. The limit of detection (LOD), defined as the lowest concentration that the analytical process can reliably differentiate from background levels, was obtained when the signal was three times the background noise in the chromatograph at the lowest analyte concentration assayed. For the limit of quantification (LOQ), the signal must be ten times the background noise. Based on the characterized ion ratio (one quantitative and one confirmation) of each compound, the intra-assay precision and accuracy were assessed at 3 levels in the range LOQ-15 µg/L. Samples out of this range were diluted to be in the range of calibration. A quality-control blank (mix of pesticide-free urine) and a quality control sample for each of the three concentration levels in the range of calibration were included every 10 samples. The calibration curve at the LOQ level was verified every 20 samples.
The calibration curves showed good linearity with a correlation coefficient >0.997. The method is precise (CV%≤20%) and accurate. Analytical characteristics resulting from the validation of the method are reported in the table below.    Urinary concentrations during pregnancy and during childhood were included simultaneously in the models.
The nonlinear component contribution was not tested in these models. The coefficients from linear models of the log-transformed exposures were thus not reported. All models were adjusted for creatinine levels of mother and child. Urinary concentrations during pregnancy and during childhood were included simultaneously in the models.
The nonlinear component contribution was not tested in these models. The coefficients from linear models of the log-transformed exposures were thus not reported. All models were adjusted for creatinine levels of mother and child. Urinary concentrations during pregnancy and during childhood were included simultaneously in the models.
The nonlinear component contribution was not tested in these models. The coefficients from linear models of the log-transformed exposures were thus not reported. All models were adjusted for HOME score, breastfeeding duration, mothers' IQ, school, maternal education level, psychologist testing the child, creatinine levels of mother and child, parity, and season of urine collection. a DAP models also adjusted for: maternal alcohol use at inclusion, and disturbances during testing b DM models also adjusted for: maternal alcohol use at inclusion, disturbances during testing, marital status, maternal fruit and vegetable consumption, maternal fish intake, and child's sex. c DE models also adjusted for: marital status, maternal fish intake, and child's sex. Urinary concentrations during pregnancy and during childhood were included simultaneously in the models.
The nonlinear component contribution was not tested in these models. The coefficients from linear models of the log-transformed exposures were thus not reported. All models were adjusted for HOME score, breastfeeding duration, mothers' IQ, school, maternal education level, psychologist testing the child, creatinine levels of mother and child. a DAP models also adjusted for: disturbances during testing b DM models also adjusted for: disturbances during testing, parity, season of urine collection, maternal fruit and vegetable consumption, and child's sex. c DE models also adjusted for: maternal fish intake.  (Engel et al. 2011) d HOME cohort study (Yolton et al. 2011; Average of the measurements of two urine samples collected at 16 weeks of gestation and 26 weeks of gestation) Figure S1. Spline regressions for studying the associations between organophosphate urinary metabolites and WISC working memory scores (n=231; PELAGIE cohort, France) Pregnancy urinary samples 6-year urinary samples DAP p-value overall association=0.40, non linear component=0.65 p-value overall association=0.84, non linear component=0.86 Models were adjusted for HOME score, breastfeeding duration, mothers' IQ, school, maternal education level, psychologist testing the child, creatinine levels of mother and child, parity, season of urine collection, maternal alcohol use at inclusion, and disturbances during testing DM p-value overall association=0.43, non linear component=0.66 p-value overall association=0.70, non linear component=0.42 Models were adjusted for HOME score, breastfeeding duration, mothers' IQ, school, maternal education level, psychologist testing the child, creatinine levels of mother and child, parity, season of urine collection, maternal alcohol use at inclusion, disturbances during testing, marital status, maternal fruit and vegetable consumption, maternal fish intake, and child's sex.