Maternal metal concentration during gestation and pediatric morbidity in children: an exploratory analysis

Exposure to heavy metals, the majority of which are highly toxic, can result in severe human morbidities. Lead (Pb), arsenic (As), and mercury (Hg) head the list, followed by aluminum (Al), cadmium (Cd), chromium (Cr), nickel (Ni), titanium (Ti), and others, whose presence is not essential to the human body. As a result, exposure to high concentrations is likely to be adversely associated with health outcomes. At the same time, other metals, including calcium (Ca), cobalt (Co), copper (Cu), iron (Fe), magnesium (Mg), manganese (Mn), molybdenum (Mo), selenium (Se), sodium (Na), and zinc (Zn), are considered essential for human physiology [1]. There is no consensus on the list of essential metals in medical literature; however, some researchers limit them to only Na, potassium (K), Mg, Ca, Zn, Se, and Cu [2, 3]. This group, also known as “biometals,” is characterized by a U-shaped association with health outcomes, whereas no U shape has been reported for non-essential metals [1].

Since toxic metals are known to cross the placenta and blood-brain barrier and deposit in fetal tissues, children are likely to be susceptible to metal exposure early, starting from the time of gestation. For this reason, maternal metal concentration is frequently used as a proxy for that of a fetus when studying health outcomes.

Maternal elevated concentrations of Al, As, Cd, Hg, Ni, and Mn have been reported in relation to newborn congenital defects at birth [4‐7]. Small-for-gestational age (SGA) status at birth has been adversely linked to Cd [8] and As [8, 9]. Likewise, preterm birth was reported associated with Cd [10, 11] and Pb [12]. Nevertheless, the association between Pb, Cd, As, Cr, or Hg and preterm birth was not always confirmed [13, 14].

A recent review by Allen provided an evidence of an association between prenatal Pb exposure and neurological and cognitive impairment [15]. The findings of several other studies support this link for Pb [16, 17], as well as for Mn [18, 19], Hg [19], As [19], and Cd [20, 21]. However, some contradictory findings have also been reported for Cd [19, 22‐24]. The lack of association in those cases might have been a result of adjustment to smoking, known to be related to Cd levels [25] as well as to cognitive function [26], and therefore a potential confounder in the association between the two.

Pulmonary function indices and asthma in children were linked with Pb [27, 28], Mn, Ni, and Cr [29, 30], as well as Co and Zn [31]. Likewise, maternal Pb has been reported in association with elevated blood pressure in 5-year-old children [32] and prenatal exposure to Cd—with an increased risk of obesity at the age of five [33].

The Bedouin-Arab population of southern Israel is known for elevated rates of childhood morbidity [34] usually related to its low socio-economic status, previous and to some degree still prevalent semi-nomadic life style, temporary housing conditions, high rates of smoking among males, and consanguineous marriages [34, 35].

Because of the vulnerability of the Bedouin-Arab community, the extensive monitoring of hazardous exposures, in an attempt to identify and eliminate modifiable factors related to morbidity, is required. In the current analysis, we aim to explore the possible links between maternal exposure to a wide range of heavy metals at gestation with birth outcomes and the health conditions of 5 to 6-year-old offspring.

In all, 111 of women and their 111 newborns had a stored maternal urine sample available for chemical analysis. There were no twins in the sub-sample. Mothers were on average (mean ± SD) 28.1 ± 6.3 years old, with more than half of them giving birth to their fourth child (Table 1). Most newborns were born at term at an average weight of 3287.3 ± 457.5 g; 10.8% were born at 37.0 weeks of gestation or earlier. The study follow-up fell within a range of 5.4-6.9 years, with  more than half of the children underwent above  6.1 years of follow-up. Women and their offspring were not different from the main cohort, as shown in its comparison to 1437/1823 women initially enrolled in the study and limited to Clalit HMO members (Table S1, supplementary materials).

Table 1 shows the list of health outcomes deemed related to environmental exposures, according to the literature. For more than half of the children, an asthma-like diagnosis was recorded in their medical charts at least once at a median age of 0.6 years, while 69.4% of all children were treated at least once for asthma-related symptoms in an emergency room or by a medication. In addition, approximately 10.8% of children at a median age of 0.8 years were referred to specialists for cardiovascular-related morbidities, although none were diagnosed with a cardiovascular disease. Over 6% of children were treated for developmental or behavioral disorders, the earliest treatment given to a 0.3-year-old child and the latest to a 6-year-old. Overall, six were referred to an outpatient developmental clinic where one child was diagnosed with developmental delay and autism. Obesity was diagnosed in 5.4% of the sample, usually around the age of 5 years or later. There were no malignancies among the 111 children in the sample, and no mortality cases were recorded. Similar to its baseline characteristics, the population of the sub-sample did not vary from its main cohort in any of the outcomes (Table S1, supplementary material).

Table 2 shows the metal concentrations in the study population. Although, based on a small sample, their distribution does not vary from the findings reported by the authors elsewhere, alongside comparison of the obtained concentrations to the values reported in NHANES population [38]. However, the comparison to a sample of 31 women in Sweden [43] and a sample of 968 women in France [44] indicated excessive levels of certain metals, e.g., Se, Al, Co, Cr, and Pb (Table 2).

Metals’ concentrations by the outcomes at study are graphically described in Figures 2 a–f (supplementary material).

Women who delivered preterm were more likely to have elevated concentrations of Zn, Tl, Al, Mn, and U with an RR of 1.40–1.64 times higher compared to in term deliveries (see Table 3 and Fig. 1 showing a heatmap of the associations presented in Table 3). Cd, Ni, and Pb were associated with preterm delivery, although with a borderline significance (0.05 < p value < 0.1) and with comparable point estimates. Asthma was linked to high concentrations of Mg, Sr, and Ba, all featured by a relatively small effect of RR close to 1.1. The behavioral and developmental group of disorders was the one most frequently associated with elevated metals, with the majority of them biometals, as well as Li, Co, Ni, Sr, Cd, V, As, and Mo. All point estimates were greater than PR = 1.42 and were different in their majority from the metals linked to preterm birth in this study.

Upon closer examination of five children with a high burden of disease (3 morbidities, etc.; see Table 1), higher concentrations were recorded for most of the metals, especially Ni and As. We did not identify any spatial clusters with respect to morbidity type or elevated metal concentrations.

We aimed to describe a possible link between maternal exposure to metals during gestation and pediatric morbidity in children aged 0 to 6 years. The vast majority of similar investigations done to this day were limited in the number of either metals tested for the main biomarker of exposure or pediatric morbidity [45, 46]. In contrast, 25 metals and all clinical conditions developed by children were at the investigators’ disposal in the current analysis.

The study population was characterized by relatively high rates of asthma-like illness, with more than half of the children diagnosed and approximately 70% of these treated for asthma. These high rates of disease are consistent with another local study in which researchers reported comparable numbers for asthma-like symptoms in the Bedouin population [47]. Moreover, the rate of malformations was high, amounting to 9.5% in the study cohort and 7.2% in the sub-sample. It should be noted that these numbers might underestimate the actual rate, since the newborns excluded from the current analysis due to antepartum or peripartum mortality were most probably those with major malformations [4, 34].

We recorded multiple links between morbidities and perinatal exposure to metals in both univariate and adjusted analyses. For instance, preterm delivery could be attributed to perinatal exposure to metals such as Tl, Al, and U as well as Cd and Pb—at significance level < 0.1. Studies of these two metals have shown contradictory results; in other words, while some studies support the link for Cd [10, 11] and for Pb [12], others do not confirm these findings [13, 14]. This inconsistency may be related to different testing matrices like a spot urine, as in case of our study, or serum [14]. These two elements are also closely related to smoking, a potential confounder of associations found. In the current study, any adjustments would not be effective, as children were exposed to second-hand smoke (from their fathers) in more than 92% of cases (data not shown).

Of note, metals linked with preterm delivery in this study were different from metals linked with behavioral and developmental outcomes, indicating some degree of specificity of exposure in its association with an outcome it may potentially produce.

Our analysis of behavioral outcomes showed a link to a slightly different set of metals, including Li, Co, Ni, Sr, V, As, Mo, and Cd. We found no previous studies that explored the association of these metals with behavioral morbidity; most focus on Cd and Pb [15‐17, 20, 21]. These together with the metals linked to preterm delivery may all represent industry-related exposure. An additional exploration of morbidity cases did not reveal any residential clusters, although our sub-sample was too small for any meaningful epidemiological investigation of an exposure source.

Despite the fact that an asthma-like illness outcome was not associated with exposure to metals with the exception of Mg, Sr, and Ba, this outcome occurred so commonly as to produce a valid analysis specific to the cause. Malformations were found to be related only to Ag, but at a borderline significance (p value < 0.1) when adjusted for other factors. The current study adds to the findings of a relatively short list of studies on the association of metal with birth defects: one shows an adverse effect of As, Pb, Cd, and Cu on neural tube defects in infants [48], another adversely links exposure to Li use in pregnancy with cardiac malformations [49], and a third, focusing on a different sub-sample of the current cohort, shows the adverse association of Al with minor malformations at birth [4]. As previously stated, the malformations we observed do not represent the complete picture of this morbidity; therefore, the current analysis is likely biased toward less severe cases.

Our findings on non-essential metals partially confirm existing theories on their U-shaped impact, i.e., whereas an element should be present within its normal range, and levels too low or too high may present a risk factor for health [1]. We recorded this type of association only for Ca and developmental outcomes, but not for other elements. At the same time, cardiovascular outcome was only negatively associated with essential metals, i.e., subjects with a cardiovascular outcome were more likely to have low levels of Zn. However, this finding was not confirmed in a multivariable analysis.

Our sample was almost identical to that of the study cohort in terms of baseline characteristics and morbidity rates. This similarity confirms the appropriateness of the subgroup’s random selection, as well as enables the extrapolation of our conclusions to a larger population.

A further limitation concerns possible misclassification based on our definitions of morbidity, where an outcome is defined by a diagnosis recorded at any point in life or treatment ever administered if deemed related to that disease. This is a frequent pitfall of definitions based on medical records, which although tending to attenuate measured associations toward the null hypothesis are not expected to bias study conclusions.

Lastly, we investigated a sample of Bedouin-Arab women and their offspring who differ in levels of hazardous exposures from the neighboring Jewish population, yet are treated by the same health services. Nevertheless, links between objectively measured metal concentrations and morbidity based on medical records rather than on self-reporting, adjusted to main risk factors, should not be limited to this ethnic group but should also hold true for other populations.

Our analysis provides many hypotheses-generating findings that warrant further exploration on a larger sample.