Nearly 1% of women aged 20–39 living in the United States have a blood lead concentration (BPb) greater than or equal to 5 μg/dL (n = 420,000) [1, 2]. Due to lead transfer across the placenta, their children are at risk of being born with BPb above the CDC reference level of 5 μg/dL [3] and are being exposed to a neurotoxicant, for which no threshold for harmful effects has been identified [4], at a period of heightened vulnerability [5, 6].

Prenatal lead exposure is linked to spontaneous abortion [7], low birth weight [8], preterm birth [9], small for gestational age birth [10], congenital malformation [8, 11], and deficits in intellectual development [12–15]. The transfer of lead across the placenta from mother to fetus is complex and incompletely understood, but is the sole factor determining fetal exposure. Sources of maternal lead can be extrinsic (maternal environmental exposure) or intrinsic (via mobilization of lead in bone stores).

The mechanisms for placental uptake and efflux of lead are not well defined, but likely include metal-specific proteins and receptors common to iron metabolism. Hence, genes that regulate iron metabolism may play an important role in placental lead transfer. Transferrin (TF) and hemochromatosis (HFE) are two iron metabolism genes for which functional variants are prevalent in North America. The TF gene is located on 3q22.1 and codes for the glycoprotein transferrin, which is responsible for intercellular iron transport [16]. While the TF gene is highly polymorphic, a common variant is the P570S missense single nucleotide polymorphism (SNP) [17]. The prevalence of the P570S polymorphism is estimated at 15% in the US general population and 12.5% in Native Americans, though the latter is based on a small sample [17].

The HFE gene is located on 6p21.3 and codes for the HFE protein, which regulates iron cellular uptake by binding transferrin and decreasing activation of transferrin receptors, thereby reducing transport across cellular membranes [18]. HFE protein influences the expression of other metal transporters found in the intestinal tract such as divalent metal transporter 1 (DMT1) [19, 20]. Two missense SNPs (H63D and C282Y) in the HFE gene are overrepresented among individuals with the hemochromatosis phenotype in which iron is hyperabsorbed. In a large screening study based in North America, 24% of whites and 20% of Native Americans carried variants for H63D, while 10% of whites and 5% of Native Americans carried variants for C282Y[21].

Though iron metabolism genes play a role in lead metabolism and toxicity among children and adults [22–29], it remains unclear how normal adaptations in homeostatic function and physiology during pregnancy may complicate the relationship between iron metabolism genes, maternal BPb, and maternal-fetal lead transfer. The objective of this study is to determine whether genomic variations in maternal and/or infant TF and HFE genes modify the association between maternal and umbilical cord blood lead.

Descriptive characteristics for mothers and infants from the original birth cohort are listed in Table 1, stratified by the availability of genotype information. Our study cohort is limited to the 476 mother-infant pairs in whom blood lead levels and genotype information was determined. The mean weight, length, and head circumference of neonates corresponded to World Health Organization international growth charts, which are recommended by the CDC for use in children up to age 24 months [31].Table 1

Characteristics of mother-infant pairs from Ottawa County, OK, stratified by availability of genotype information ^a

Genotype frequencies for all 476 mother-infant pairs were successfully determined for all three SNPs. The genotype frequencies for mothers and infants are shown in Tables 2 and 3, respectively; all were found to be in Hardy-Weinberg equilibrium (α = 0.05). There were no statistically significant differences between geometric mean blood lead values when stratified by variant status.Table 2

Maternal genotype frequencies and geometric mean blood lead levels for dominant models ^a,b

In a bivariate analysis restricted to wild-type mother-infant pairs (n = 157), MBL was linearly associated with UCBL (β = 0.79; 95% CI 0.69, 0.88; p < 0.001). When this analysis was repeated for the log-transformed values, the correlation between these two variables strengthened (β = 0.99; 95% CI 0.88, 1.10; p < 0.001). Similar parameter estimates were obtained when the bivariate analysis was expanded to include all 476 mother-infant dyads (data not shown).

In our sensitivity analysis, the association between maternal and umbilical cord blood lead was not confounded by gestational age, birth weight, infant gender, maternal race/ethnicity, socioeconomic status, and infant hematocrit or serum ferritin. While the parameter estimate for maternal hematocrit at delivery was statistically significant (β = −0.008; 95% CI −0.01, −0.003; p < 0.01), it did not meaningfully change the beta estimate for MBL (β_adjusted = 0.84; 95% CI 0.77, 0.92; p < 0.001; β_unadjusted = 0.85; 95% CI 0.78, 0.93; p < 0.001). Therefore, it did not satisfy the a priori definition for confounding.

Lead-hematocrit interaction terms were also modeled under the premise that hematocrit might modify placental lead transfer if, for example, varying amounts of lead cross the placenta in the presence of low or normal hematocrit. The interaction between MBL and both maternal hematocrit at delivery (β_Interaction = −0.02; 95% CI −0.04, 0.001; p = 0.06), as well as maternal hematocrit at 28 weeks (β_Interaction = −0.02; 95% CI −0.05, 0.003; p = 0.08) were marginally significant. While the interaction between MBL and umbilical cord hematocrit at birth (β_Interaction = −0.03; 95% CI −0.04, −0.01; p = 0.001) was statistically significant, this result should be interpreted with caution; since umbilical cord hematocrit was not routinely measured, the sample available for analysis was limited to only 132 of the original 476 study participants.

Lastly, we explored the potential for a gene-environment interaction between iron metabolism genotype and MBL by regressing log-transformed values of UCBL on similarly-transformed values of MBL. In order to account for potential confounding between genotype variants, the main effect terms for maternal and infant TF and HFE variant status, along with their respective interaction with MBL, were modeled together. Though infant and maternal variant status were modestly correlated for each gene variant (Spearman ρ ranging from 0.43 to 0.52), variance inflation factors did not exceed 3.4 and thus raised little suspicion for collinearity. This conclusion was strengthened by two additional observations: 1) parameter estimates remained stable when potentially collinear variables were selectively excluded from the full model described in Table 4 and 2) in regression models that were limited to either infant or maternal gene variants (Additional files 1 and 2, respectively), the main effect and interaction term parameter estimates were very similar to those from the full regression model.Table 4

Multivariable regression model of iron metabolism gene variants, log-transformed maternal blood lead, and their interaction on log-transformed umbilical cord blood lead ( n =  476) ^a,b,c

In a fully-adjusted model accounting for infant and maternal iron metabolism genotype and gene-environment interaction terms, MBL retained its association with UCBL (β = 0.92; 95% CI 0.82, 1.03; p < 0.001; Table 4). Among wild-type mother-infant pairs, a 1% increase in MBL is expected to increase UCBL by 0.92%. The association between UCBL and MBL was not modified by infant gene variant status or infant genotype-MBL interaction terms (Figure 1). Likewise, maternal gene variant main effect terms were not significant. However, the interaction between maternal HFE C282Y genotype and MBL was significant (β_Interaction = −0.27; 95% CI −0.52, −0.01; p = 0.04; Table 4) and attenuated the relationship between UCBL and MBL (Figure 2). While the other two maternal genotype interaction terms were not statistically significant, all three consistently trended towards a protective effect on UCBL (Figure 3). Notably, these results did not differ appreciably from output generated by the two regression models that were limited to either infant or maternal gene variants (see Additional files 1 and 2, respectively).Figure 1

Modification by infant genotype of the association between maternal and umbilical cord blood lead. Adjusted for maternal genotype. Three values ≤0.005 ug/dL for umbilical lead not shown.

The results of this study suggest that maternal iron metabolism genotype is a modifier of placental lead transfer. Infants born to women with the HFE C282Y gene variant have lower umbilical cord blood lead concentrations. Further, the attenuating impact of this gene-environment interaction on placental lead transfer strengthens as MBL rises. While MBL seems to be the key determinant of UCBL, maternal iron metabolism gene variants modify this relationship in a manner that is potentially clinically significant in exposed populations.

Additional results suggest that fetal iron metabolism gene variant status did not modify the relationship between UCBL and MBL. In other words, it appears that lead transfer across the placenta is not altered significantly by fetal iron metabolism genes, but primarily by maternal genotype. This finding has direct relevance for explaining the biology that underlies placental lead transfer.

At least four polymorphic genes are recognized to influence lead toxicokinetics and toxicodynamics: delta-aminolevulinic acid dehydratase (ALAD), vitamin D receptor (VDR), TF and HFE[22, 32]. Genetic variability may influence bioavailability and susceptibility in a complex fashion, interrelating with other individual characteristics such as age and micronutrient status to produce differential outcomes. For example, studies have shown that while elderly men who carry HFE gene variants have lower tibia, patella, and blood lead concentrations than wild-type subjects [23], children who carry either HFE or TF variants have higher BPb than wild-type peers [22]. Moreover, the impact of iron metabolism gene variants on lead metabolism seems to be amplified in the presence of multiple genotypic variant alleles, akin to a dose–response relationship [22, 24, 25].

While the role of iron metabolism gene variants in primary lead exposure and bioaccumulation has been studied, relatively little is known about how maternal and fetal genetic polymorphisms affect lead transfer across the placenta. This study adds to the existing literature by providing evidence that fetal lead exposure is attenuated among children born to mothers who carry variants for maternal iron metabolism genes.

The strong association between umbilical cord and maternal blood lead seen in this cohort has been observed in many studies [33–36]. However, the toxicokinetics of placental lead transfer are not well understood. Lead uptake by the placenta occurs passively [37] and is likely influenced by several proteins including, for example: HFE protein, which is expressed on the apical plasma membrane of placental syncytiotrophoblasts near the maternal circulation, and which interacts with transferrin to regulate cellular iron absorption [38]; DMT1, an iron transport protein that is also found in syncytiotrophoblasts and other tissues, and that plays a vital role in iron homeostasis [39–42]; and metallothioneins, which are small, cysteine-rich proteins found in abundance in placental tissue, have high metal binding properties, and participate in metal storage and detoxification [43–45]. Less is known about mechanisms underlying lead efflux [5]. Though the basolateral iron exporter ferroportin seems to be the sole cellular iron transporter [46], it may or may not play a role in lead transport.

Placental iron transfer occurs at the apical plasma membrane of syncytiotrophoblasts via transferrin receptor-mediated endocytosis. Since HFE protein is expressed at this site and is physically associated with the transferrin receptor, it has been hypothesized that HFE protein regulates the transfer of iron from maternal to fetal circulation [38]. While wild-type HFE protein binds to β₂ microglobulin and is transported to the cell surface where it interacts with transferrin to regulate iron, these normal functions are altered in individuals who carry the HFE C282Y gene mutation. Given that lead is carried by these same iron transporters and receptors, it is possible that functional perturbations in the HFE protein resulting from HFE C282Y gene mutations may modify placental lead transfer.

Other hypotheses may also explain the biologic basis for decreased placental lead transfer among women who carry iron metabolism gene variants. For example, these women are more likely to have heightened pre-pregnancy stores of iron, which may decrease expression of metal-specific transporters or proteins and thereby lessen gastrointestinal lead absorption. Another potential explanation is the creation of a lead-sink in which lead moves freely into placental tissues via up-regulation of metal transport mechanisms [26] but is slow to exit into the fetal circulation due to changes in expression of proteins that bind lead. Perhaps by saturating these transporters, higher maternal levels of micronutrients – specifically zinc and manganese –reduce lead transfer to the fetus [34].

Other research in this area has shown that genetic polymorphisms in iron metabolism genes not only modify lead kinetics, but also confer variable host response to lead exposure. Carriers of the TF P570S gene variant, for example, lose significantly more IQ points per unit of blood lead than children with wild-type genotypes [27]. Lead-related decreases in birth weight [28], advancement of cognitive decline [24], widening of pulse pressure [29], and prolongation of the QT-interval [25] have been shown to be amplified in HFE gene variant carriers.

This study has several limitations. Whereas lead toxicokinetics are likely influenced by an array of functional gene polymorphisms, such as those involved in calcium regulation, we investigated only a few iron-specific candidate genes. Other genetic variants may modify placental transfer as well, but we did not have sufficient information in our dataset to examine them. Additionally, genes in linkage disequilibrium with the HFE gene might explain our results. Such confounding is unlikely, however, because other known iron transport genes are not located in this genomic region and HFE C282Y is a well-characterized functional variant.

Though it is characteristic of the US population, the narrow range of exposure among women in this study limits our ability to extrapolate findings to populations that have higher average blood lead levels. While residual confounding is possible, any environmental factor that confounded our findings would have to be associated with both placental lead transfer and HFE genotype. Apart from micronutrient status of iron and calcium, for example, it seems unlikely that there are environmental factors that could be associated with both placental lead transfer and HFE genotype. Introducing additional biomarkers of iron and calcium sufficiency might clarify the influence of maternal and fetal micronutrient status on placental lead transfer.

Statistical considerations include the possibility that the significant interaction between MBL and maternal HFE C282Y may be due to a Type I error related to multiple testing. Conversely, a Type II error due to insufficient power may explain why the remaining gene-environment interaction parameter estimates were no different from the null. Finally, like all new findings, these results require validation in other cohorts.

When taken in the context of current trends in clinical medicine and population health, the results of this study have interesting implications. Identifying factors that modify placental lead transfer might inform research in the emerging era of genomic medicine, where targeted therapeutics hold the promise of lessening the morbidity associated with prenatal lead exposure. Information about genetic modifiers of lead transfer may also inform how clinicians will advise, treat, and monitor their pregnant patients for micronutrient deficiencies. From a regulatory standpoint, results of this study prompt us to reconsider the current approach to risk assessment and ask whether methods should consistently address issues of genetic susceptibility.

This study is among the first to investigate how maternal and fetal genetic polymorphisms secondarily affect lead transfer across the placenta. Through gene-environment interaction, maternal iron metabolism genotype appears to play a role in maternal-fetal lead transfer. We found that infants born to mothers with the HFE C282Y gene variant had lower umbilical cord blood lead levels relative to those born to women who were wild-type. Moreover, reductions in the proportion of maternal blood lead transferred across the placenta were greatest for infants whose mothers had higher blood lead levels. The incorporation of gene-environment information into risk assessment may improve regulatory efforts to safeguard the health of vulnerable populations.