Maternal organophosphate pesticide exposure in relation to birthweight: Modification by placenta transporter genotype

Study sample

The sample included 240 participants enrolled in the New York University Children’s Health and Environment Study (NYU CHES), an ongoing prospective birth cohort beginning in 2016 in New York City [21]. Sociodemographic, lifestyle, and clinical characteristics were obtained via self-reported questionnaires during prenatal visits or abstracted from medical records. Sex-specific birthweight-for-gestational-age (BWGA) z-scores were calculated using the INTERGROWTH-21^st standard [22]. Participants were selected for this ancillary study if they had previously generated data on prenatal OPP exposure and candidate placenta transporter genotype (Supplemental Figure 1). The analytic sample had similar sociodemographic and birth outcome data compared to the overall sample of singleton, livebirths enrolled in the cohort (Supplemental Table 1).

Maternal OPP exposure during pregnancy

The majority of OPPs are biotransformed into six non-specific dialkyl phosphate (DAP) metabolites (dimethylphosphate [DMP], dimethylthiophosphate [DMTP], dimethyldithiophosphate [DMDTP], diethylphosphate [DEP], diethylthiophosphate [DETP], and diethyldithiophosphate [DEDTP]) [23]. We measured their concentrations in maternal urine collected at up to three gestational time points using high-performance liquid chromatography (HPLC) coupled with electrospray triple quadrupole mass spectrometry (ESI-MS/MS) [3]. Due to low detection (11%), DEDTP was excluded from inferential analyses. We replaced non-detectable values (Supplemental Table 2) of the other metabolites with the limit of detection divided by the square root of two. We also replaced one extreme DMP and DMDTP outlier with the next highest value in the distribution to retain these datapoints while minimizing their influence in regression analyses. DEP and DMP had the highest concentrations, with median values of 2.64 and 1.92 ng/ml, respectively. DMTP followed with a median concentration of 1.48 ng/ml, while DETP and DMDTP had lower concentrations (medians 0.34, 0.04 ng/ml). We calculated the molar sum of dimethyl (∑DM: DMP, DMTP, DMDTP), diethyl (∑DE: DEP, DETP), and all five DAP (∑DAP) metabolites, which we subsequently corrected for urine dilution using creatinine following the Boeniger approach as previously described [3,24]. Molar sums were then log transformed and averaged across pregnancy. Because DAPs have a short half-life in humans [2], averaging across pregnancy provides a more representative index of exposure, yet we acknowledge that concentrations in a spot urine sample and fluctuate and may not accurately represent long-term exposure.

Placenta transporter SNPs

Placenta samples were collected at delivery, flash frozen in liquid nitrogen, and stored at −80°C until genotyping by the Johns Hopkins Nucleic Acid Technologies Lab (RRID:SCR_018669). Candidate SNPs included: rs2231142 ABCG2/BCRP; rs3213619 ABCB1/MDR1; rs12422149 SLCO2B1/OATP2B1; and rs2078267 and rs17300741 for SLC22A11/OAT4. Detailed methods are provided in supplemental material. Briefly, genomic DNA was isolated before end-point real-time PCR for DNA sequence detection. Given low variability in genotype, the rs3213619 SNP was not further considered. Figure 1 depicts the genotype frequencies for the four SNPs. We assessed Hardy-Weinberg equilibrium of SNPs using population stratification based on self-reported race/ethnicity. All four SNPs were in equilibrium with the exception of two among those who identified as Hispanic, which we highlight is a heterogenous group that includes genetically diverse individuals (Supplemental Table 3). Genotype was not independently associated with BWGA in the sample (Supplemental Table 4).

Statistical analysis

We descriptively evaluated the data using summary statistics, boxplots, and histograms. We investigated overall unadjusted and adjusted associations between each OPP molar sum and BWGA z-score using separate linear regression models. Adjusted models included maternal age, education level, race/ethnicity, insurance status, pre-pregnancy body mass index, alcohol consumption during pregnancy, delivery method, and infant sex. Next, we included a cross-product term to evaluate the interaction between each OPP molar sum and candidate transporter SNP, treating the wild-type allele as reference. We additionally ran models stratified by genotype to visualize associations using forest plots (Figure 1). All statistical analyses were conducted using R version 4.2.1.

Study sample characteristics

Among the 240 participants, 134 (55.8%) identified as Hispanic, 58 (24.2%) as non-Hispanic White, 17 (7.1%) as non-Hispanic Black, and 31 (12.9%) as another racial/ethnic group. The mean maternal age was 31.0 years (SD=5.8) and 94 (39%) had a high school or lower education level. Additional characteristics are presented in Table 1.

OPP – BWGA associations

As hypothesized, ∑DAP, ∑DM, and ∑DE were inversely related to newborn BWGA, although associations did not reach statistical significance. When examining stratified models, all three molar sums were significantly inversely associated with BWGA for OATP2B1 rs12422149 A/G, OAT4 rs2078267 C/T and OAT4 rs17300741 A/G SNP variants. This was observed in heterozygous individuals whereas no association was detected in the smaller sample of individuals with two variant alleles. For the OAT4 rs17300741 G/G variant, we observed a positive effect estimate with wide confidence intervals (CIs) in stratified models; however, we interpret this result cautiously given the low number of participants (n=28). We did not find evidence that BCRP genotype modified associations between maternal OPP exposure and BWGA. Results are depicted in Figure 1 with beta coefficients and 95% CIs provided in Supplemental Table 5.

Placenta transporters have been implicated as gatekeepers that regulate fetal exposure to pharmaceuticals and environmental toxicants [25–29]. This preliminary study is the first to consider whether they may influence fetal exposure to OPP metabolites. Our findings suggest that pregnancies with one copy of variant alleles in SNPs for OATP2B1 and OAT4 may be more vulnerable to OPP exposure due to reduced clearance of OPPs from the fetal compartment. While it is perplexing that we did not observe corresponding significant associations among individuals with two variant alleles, we emphasize the smaller sample size and limited power to detect associations in this group. Our findings support the need for future larger studies that are designed to investigate transporter gene by environment interactions in the placenta. We also highlight that while we selected OATP2B1 and OAT4 as candidates based on their abundant expression by the placenta and ability to transport xenobiotics [30,31], additional in vitro research is needed to test individual OPPs as substrates. While this study is novel in scope, we acknowledge limitations including a modest sample size, no measurement of placenta transporter proteins, potential maternal contamination of placenta DNA, and no direct measure of fetal OPP concentrations. Future, larger studies are needed to confirm these preliminary findings and test interactions of OATP2B1 and OAT4 with OPP metabolites with respect to other perinatal outcomes. In summary, this study highlights the role of placental genetic variability in modifying relationships between environmental xenobiotic exposures and fetal growth. Our findings support a recent call by the International Transporter Consortium for the urgent consideration of placental transporter variants in ongoing birth cohort studies as a novel mechanism to interrogate transporters as regulators of the interaction between perinatal exposures and postnatal outcomes [32].