The participants of this study are part of the PROGRESS (Programming Research in Obesity, Growth, Environment and Social Stressors) prenatal cohort from Mexico City. Pregnant women receiving medical and prenatal care in the Mexican Social Security Institute (IMSS) were invited to participate. The recruitment period took place between July 2007 and February 2011. To be eligible for the study, women had to be at <20 weeks gestation, ≥18 years of age, without current or prior pathologies of heart or kidney disease, accessibility to a phone, plan to reside in Mexico City for the next 3 years after their admission to the study, not use steroid drugs (including glucocorticoids) or antiepileptic medications, and not consume alcohol daily. The institutional boards and ethics committees of Harvard School of Public Health, Icahn School of Medicine in Mount Sinai, and the National Institute of Public Health in Mexico approved the project, as well as the collaborating institutions: National Institute of Perinatology (INPer), IMSS, and the American British Cowdray (ABC) Medical Center. Participants granted their informed consent for participation in the study. There were 948 women followed during pregnancy and who gave birth to a live child; children had follow-up visits at 1, 6, 12, 18, 24, 48, and 72, and 96 months of age. The cohort had most of its dropouts between birth and at the stage 24 visit and has maintained stable retention at around n = 600 mother–child dyads since.

The inclusion criteria for this study were: having data of at least one blood lead level (BLL) during pregnancy and one postpartum (i.e., second and third trimesters of pregnancy, birth, and at 12, 24, and 48 months of age) and outcome data (i.e. MetS indicators) at 72 and 96 months. Some of the children's blood samples from stages 12 and 24 were randomly lost during transportation from Mexico to the laboratory in the USA, leaving a total of n = 174 and n = 247, respectively. For stage 48, we have the data for 501 children. Children who had an extremely low birth weight equivalent to <1,500 g and gestational age <32 weeks were excluded (n = 21). On average, 601 children had data for at least one incidence of MetS for stage 72 and 540 had data for stage 96, with varying covariate data during the study follow-up that are indicated in the corresponding results table.

Prenatal exposure was evaluated with blood measurements from women in the second and third trimesters of pregnancy and in umbilical cord blood. Postnatal Pb exposure was assessed from children's blood samples collected at 12, 24, and 48 months of age. All blood samples were drawn in trace metal-free tubes and refrigerated at 4°C until shipment to the laboratory where they were frozen at −20°C until analysis. Pb concentration was measured by external calibration using the Agilent 8800 ICP Triple Quad (Agilent, Santa Clara, CA, USA) in MS/MS mode in the Trace Metals Laboratory at the Icahn School of Medicine at Mount Sinai. The limit of detection was <0.2 μg/dl and the instrument precision (given as %RSD) was ~5%. Good precision and accuracy were shown using blinded quality control samples obtained from the Maternal and Child Health Bureau and the Wisconsin State Laboratory of Hygiene Cooperative Blood Lead Proficiency Testing Program. BLLs were analyzed as continuous variables (given in micrograms per deciliter) and as categorical variables classified according to the median exposure per stage, described below.

To account for cumulative exposure, at the prenatal, postnatal, and cumulative stages, we dichotomized the BLLs using the median for each study visit as the cutoff point. Two categories were obtained: “higher,” when a BLL was at least one time above the median (HPb), and “lower,” if all BLLs were below the median (LPb).

We measured bone Pb levels at 1 month postpartum in maternal tibia (mid-tibial shaft, cortical bone) and patella (trabecular bone) using a K-shell X-ray fluorescence instrument for 30 min in each leg. The measures were computed, averaged, and weighted by the inverse of the proportion of the measurement error corresponding to each determination. Negative values, produced when the true values are below the detection limit of the instrument, were imputed with random draws from a uniform distribution between 0 and the lower limit of 2 μg lead/g bone mineral (23).

Metabolic syndrome risk factors (IMetS), which were assessed at the 72- and 96-month visits, included the following: glucose, glycated hemoglobin (HbA1c), total cholesterol (TC), triglycerides (TGs), high-density lipoprotein cholesterol (cHDL), low-density lipoprotein cholesterol (cLDL), body fat percentage (BF%), waist circumference (WC), body mass index (BMI), systolic blood pressure (SPB), and diastolic blood pressure (DBP).

Fasting venous blood was drawn to obtain blood levels of TC, cHDL, cLDL, TGs, glucose, and HbA1c. Enzyme methods (Roche Diagnostics, Indianapolis, IN, USA) were used to obtain measures of TC, cHDL, cLDL, and TGs. The levels of HbA1c were determined using the Miura 200 automated analyzer (ISE S.r.l., Rome, Italy). The weight, height, and waist circumference were measured during both follow-ups in triplicate and the average of each measurement obtained. The weight and height data were used to calculate the BMI (weight/size²) and the z-score with BMI for age indicator using the WHO “Anthro” software (24). Total adipose tissue was measured using the InBody 370 or 230 tetrapolar bioelectric impedance equipment (Biospace Co., Ltd., Los Angeles, CA, USA). BF% was obtained by estimating the total adipose tissue divided by the total body mass. WC was measured at the midpoint between the last floating rib and the iliac crest. Standardized personnel performed all anthropometric measurements.

SBP and DBP were measured at rest using an automated Spacelabs Healthcare monitor, Ultralite Ambulatory 90217 (Snoqualmie, WA, USA). This measurement was taken in duplicate with a 3-min difference between each, and the average of the measurements was used.

IMetS were categorized according to the different cutoff points (24, 25) considering high levels in blood if: glucose was ≥100 mg/dl, HbA1c was ≥5.7%, TC was ≥170 mg/dl, cHDL was <45 mg/dl, cLDL was >110 mg/dl, and TGs were >75 mg/dl for children 9 years of age or younger and >90 mg/dl for children 10 years of age or older. The cutoff point for BF% and WC was the 80th percentile. BMI z-scores were classified according to the WHO Child Growth Standards (24): underweight, less than −2 SD; normal weight, greater than −2 to less than +1 SD; overweight, greater than +1 SD; and obese, greater than +2 SD. The cutoff point for SPB and DBP was the 90th percentile.

The following variables were considered for model adjustment: maternal age at recruitment, socioeconomic status (SES), parity collected through standardized questionnaires during the second trimester of pregnancy, child's size for gestational age (Fenton score), sex, and age at the 72- and 96-month visits. SES was measured using the Mexican Association of Market Intelligence and Opinion Agencies questionnaire. With reference to our study population, which belongs to a middle–lower SES, we collapsed the variable into three categories: lower, medium, and higher. Parity considered two categories (including the current pregnancy): primiparous (less than or one pregnancy) and two or more pregnancies.

Descriptive analyses were carried out for the evaluation of extreme or implausible values in the database. In the case of continuous variables, the mean and standard deviations were obtained, as well as the ranges or the median and interquartile range, as appropriate. Frequencies and percentages were obtained for categorical variables. Student's t-test and the Mann–Whitney U test were used for continuous variables and the chi-square test used for categorical variables.

To assess the association between exposure to Pb and IMetS, linear mixed effects models and logistic mixed effects models were utilized. BLLs were evaluated both as continuous and dichotomous. IMetS were assessed as continuous and categorized. The models were adjusted for sex, maternal and child ages, parity, SES, and the Fenton score. Secondary analyses were additionally adjusted for the BMI z-scores to ensure that obesity did not impact the results. Models were utilized considering three developmental windows: prenatal, postnatal, and cumulative. We stratified our models by sex to assess possible effect modifications of the associations. Outliers of the IMetS were excluded (n = 4), three with biologically implausible BF% data and one with a BMI z-score for age ≥5 SD. All analyses were performed using Stata version 14 software (StataCorp LLC, College Station, TX, USA).

As a sensitivity analysis, mixed effects models were utilized with a subsample that had the complete follow-up data to verify that loss to follow-up and the presence of missing values did not affect the results.

Our study suggests that HPb, both postnatally and during all stages, was associated with higher odds of having lower TC levels. This result could be explained by the effect of prenatal exposure to Pb in the regulation of cholesterol metabolism (20, 28). According to the study by Liu et al., exposure to Pb was inversely associated with the TC levels in boys (20), and the direction of the association was similar to what we have observed in the present study. Kupsco et al. found no association between the BLLs in pregnancy and TC in 4- to 6-year-old children (17). In the study by Poursafa et al. (21), positive associations were found across quartiles of the BLLs and TC in children and among girls.

In our study children with higher levels of Pb in the patella were more likely to have cHDL levels <45 mg/dl. These results were similar to prior evidence, which showed that prenatal BLLs were inversely associated with cHDL at age 10–18 years (20) although a null association was reported by Poursafa et al. (21) in Iranian pre- and adolescents.

Prenatal BLLs and tibia Pb levels were inversely associated with TGs. Most of the evidence has reported null associations (17, 20), yet one study on Iranian boys and girls found that an increase in the quartiles of BLLs was positively associated with TGs (21). In our study, concordance between the direction of the associations between BLLs in the prenatal stage and the cumulative Pb levels measured in the tibia may suggest that exposure during pregnancy could impact the TGs levels in 6- to 12-year-old children.

Not surprisingly, we observed that an increased prevalence of overweight and obesity added to the increased WC measures and BF% at the stage 96 study visit, which could be due to changes in body composition associated with age (29).

Similar to what have been reported, we found an inverse association between HPb and BMI for age (30–34). This evidence shows that both prenatal (32, 34) and postnatal (33) exposures to Pb adversely impact children's growth. In the study by Shao et al., negative associations were found between urinary levels of Pb and overweight and obesity, being stronger among 6- to 12-year-olds compared to those in 13- to 19-year-olds (30). In the study by Liu et al., Pb levels in the patella were negatively associated with lower child BMI z-scores (β = −0.02, 95%CI = 0.03 to −0.01). In prior studies of the PROGRESS cohort, Renzetti et al. (Pb during the last two trimesters of pregnancy, delivery and postpartum, with BMI in children at 4–6 years) (32) and Kupsco et al. (BLLs during the third trimester of pregnancy and the BMI z-scores at 48 months) (17) reported null associations. Similarly, null associations were found by Afeiche et al. between longitudinal exposure to Pb and BMI in 48-month-old Mexican children. However, other findings showed that children with high exposure in all windows of development (prenatal, infancy, and early infancy) were almost 1 cm (β = −0.98, 95%CI = −1.86 to −0.10) shorter than children with low exposure during those periods, in birth cohorts in Mexico City (35).

Although the evidence is not conclusive, we propose three main biological mechanisms that could be involved. Firstly, the impact of Pb exposure in utero resulting in lower weight (36, 37) and height at birth (37): this effect can continue during early childhood (38). A second mechanism is the action of Pb as an endocrine disruptor by reducing the response of hormones such as insulin like-growth factor (39) and the action of cortisol that could affect the hypothalamus–pituitary–adrenal axis (40). The third and most reported mechanism involves the main effect of Pb on bone growth by impairing the function of bone cells (osteoblasts), the mineralization of bone, and the ability of the bone to respond to hormonal regulation, which can also affect the bone cartilage (41, 42).

We found a similar association between HPb and WC to that reported in the study by Liu et al., where the Pb levels in the patella were inversely associated with WC (β = −0.12 cm, 95%CI = 0.22 to −0.03) and BF% (β = −0.09%, 95%CI = 0.17 to −0.01) (34). Similar associations were also reported by Deierlein et al. in a study of girls (33). In the case of BF%, null associations were found in our study, similar to results previously reported in the PROGRESS cohort with children in other developmental stages (17, 32).

Contrary to what we expected, our results showed that HPb was associated with less likelihood to have elevated SBP and DBP (above the 90th percentile). Positive associations have been shown between Pb exposure and SBP using different biomarkers such as maternal toenail (27), maternal tibia bone (26), and children's BLLs (21). Null associations between maternal exposure to Pb and blood pressure were reported by Skröder et al. in a cohort of 4.5-year-olds rural in Bangladesh (43) and by Kupsco et al. in the same study population as ours, children 4–6 years old (17). Poursafa et al. showed positive associations between the quartiles of exposure to Pb and SBP and DBP. More studies are needed to explain the biological mechanisms that could produce these changes on vascular systems.

We are aware of the possibility of a type 1 error due to multiple comparisons; however, after Bonferroni adjustments, the presented associations between HPb and TC were still statistically significant. One of our study limitations is having missing BLLs in the 12- and 24-month postnatal stages, but the results in the models indicated that these losses did not influence the associations found.

Exposure to lead continues to be a public health problem in Mexico since the main source of exposure is the use of lead-glazed low-temperature ceramics to prepare, serve, and store food. Lead will leach into food with each use, even in very old and worn-out dishes. This type of ceramics is widespread in Mexico, and recent studies using representative national data have shown a clear association between their use and BLLs in 1- to 4-year-old children (44, 45).

The strengths of this study include it being the first epidemiological study that evaluated the association between repeated measures of exposure to Pb (separately during prenatal, postnatal, and cumulative early life stages) and the risk of IMetS longitudinally at two different time points, between ages 6 and 12 years. A particular strength is our extensive data on Pb exposure in both blood (acute exposure, with a half-life of 25 days) and bone [indicator of chronic exposure (half-life of decades) and endogenous exposure during pregnancy and long-term exposure in mothers]. The PROGRESS cohort is a prospective study that collected data on 11 risk factors of IMetS evaluated at two different ages, as well as covariates for model adjustment.