Correlates of prenatal heavy metal contamination in Lebanese infants using cord blood analysis

Abstract

Exposure to heavy metals during pregnancy can disrupt fetal development and cause adverse health effects. However, data on fetal exposure in Lebanon remain scarce. The Environmental Exposures in Lebanese Infants (EELI) Study, a longitudinal birth cohort launched in 2021, investigates early-life environmental exposures among Lebanese infants. This study assesses the levels of 8 toxic (antimony, arsenic, chromium, nickel, cadmium, mercury, thallium, and lead) and 6 essential heavy metals (molybdenum, manganese, cobalt, copper, zinc, and selenium) in cord blood samples (N = 74). Descriptive statistics were used to characterize metal concentrations, and bivariate and multivariate analyses were conducted to examine associations with selected sociodemographic, environmental, health-related, and dietary factors. All toxic heavy metals were detected in cord blood samples, indicating prenatal exposure. Lead (md: 5.45 µg/L, p95: 22.4 µg/L) and mercury (md: 0.7 µg/L, p95: 2.01 µg/L) were among the highest detected toxicants. Essential metals such as zinc, copper, and manganese exhibited the highest variability. Several contextual factors, including residential proximity to highways, parking lots, and industrial areas, home renovations, energy sources, and pesticide use, were statistically associated with variations in cord blood metal concentrations. While causal inference cannot be established, this study underscores the importance of documenting early-life exposure patterns to inform future research into the developmental origin of health and diseases in resource-limited settings.

Introduction

It is widely recognized that the period of human development preceding birth constitutes a critical and delicate stage of human development due to accelerated fetal cellular division and differentiation¹. During this period, exposure to toxic substances can have profound and long-lasting consequences, affecting fetal growth and increasing the risk of adverse health outcomes in early childhood and later life². The Developmental Origins of Health and Diseases (DOHaD) hypothesis underscores the long-term effects of maternal exposures during pregnancy on offspring health and highlights the central role of environmental factors in determining disease susceptibility³. Moreover, the placenta, a unique, temporary organ and a vital interface between the mother and the developing fetus, plays a pivotal role in regulating nutrient transport and filtering potential toxins^4,5. Despite being considered as a protective “gatekeeper,” research has demonstrated that several toxicants can either partially or fully bypass this barrier and reach fetal circulation^6,7 Environmental exposure to such toxicants typically occurs through ingestion, inhalation, and dermal contact; once internalized, these contaminants may circulate in maternal blood, where pregnancy-related physiological changes such as increased blood volume, altered renal clearance, and mobilization of mineral stores, can influence distribution and bioavailability^1,6. This is problematic as this interference can alter fetal programming and expose the fetus to environmental contaminants from air, water, soil and/or consumer products and lead to adverse pregnancy outcomes.

Heavy metals are a heterogeneous group of elements characterized by their high density (> 5 g/cm³) and persistence in the environment⁸ These elements can be classified into two broad categories: (1) toxic heavy metals, including Antimony (Sb), Arsenic (As), Chromium (Cr), Nickel (Ni), Cadmium (Cd), Mercury (Hg), Thallium (Tl), and Lead (Pb), which exhibit toxic effects even at low concentrations⁹ and (2) essential heavy metals, including Molybdenum (Mo), Manganese (Mn), Cobalt (Co), Copper (Cu), Zinc (Zn), and Selenium (Se), which are vital for physiological functions but can become harmful in excess^8,10. Heavy metals have been detected in various biological matrices of pregnant women, including urine, peripheral blood, and amniotic fluid, raising concerns about their toxicity, particularly during fetal development¹¹. Several toxic and essential metals have been shown to cross the placental barrier via passive diffusion, active transport mechanisms, or metal-binding proteins, resulting in direct fetal exposure⁸.

Fetuses exhibit heightened vulnerability to the toxic effects of heavy metals due to their underdeveloped detoxification systems and rapid organ development. Even at low levels that are unlikely to affect mothers, neural toxicity, bone accumulation, enzymatic inhibition, and structural and/or functional abnormality can occur and affect the infants, depending on the dose, the timing, and the duration of the exposure¹¹. For several toxic heavy metals, no clearly established safe exposure threshold during pregnancy has been identified, as these elements can cross the placental barrier, bioaccumulate, and interfere with critical developmental processes. In contrast, essential trace elements are required for normal development but may become harmful at deficient or excessive levels. Furthermore, the physiological changes that occur during pregnancy, such as increased blood volume, renal perfusion, and hormonal fluctuations, can alter heavy metal absorption, distribution, and excretion¹². The maternal heavy metal burden is also influenced by socio-demographic and lifestyle factors, including socioeconomic status, housing conditions, diet, and smoking exposure¹⁰, underscoring the complexity of environmental influences on pregnancy outcomes.

In many low- and middle-income settings, including Lebanon, pregnant women may experience exposure to heavy metals through a combination of environmental and lifestyle-related contexts. These include residence in densely populated urban areas with high traffic density, proximity to informal or small-scale industrial activities, aging housing infrastructure, and reliance on private power generators or solid fuels for electricity. These conditions are particularly prevalent in urban and peri-urban regions of Lebanon, where residential, commercial, and industrial land uses often overlap and environmental regulation and monitoring are limited. In addition, agricultural practices and household pesticide use may contribute to exposure through contaminated food, soil, or indoor environments. Such contexts may place vulnerable populations, including pregnant women and young children, at increased risk of cumulative exposure to heavy metals.

Research conducted in Lebanon has examined the presence of heavy metals in various biological matrices and their associations with various public health outcomes. Studies have assessed the levels of heavy metals in breast milk¹³, dietary intake^14,15, and particular foods like canned foods¹⁶ local herbs¹⁷, and fish products¹⁸, putting into evidence exposure to lead, cadmium, arsenic and mercury. Despite the growing public health concern about exposure to heavy metals during pregnancy, data concerning the levels of these elements in cord blood remain limited in Lebanon. Additionally, previous studies have only assessed blood lead levels in children¹⁹, cadmium contamination in water sources²⁰ and occupational/environmental exposure in relation to male infertility^21,22. While these studies provide critical insights, no study to date has examined the burden of heavy metals in cord blood, leaving a gap in understanding direct fetal exposure. Biological monitoring is a widely used approach for assessing internal exposure to heavy metals in humans, as it integrates multiple exposure routes and reflects the biologically relevant dose. Among available matrices, cord blood is considered particularly suitable for evaluating prenatal exposure, as it directly reflects fetal circulation at birth and captures metals transferred across the placenta during gestation. To the best of our knowledge, this is the first study in Lebanon that aims to comprehensively assess a wide range of heavy metals in cord blood in Lebanon, providing valuable data on infants’ exposure and its potential sociodemographic and environmental correlates.

Results

Population characteristics

The mean age of participants is 31.9 ± 3.91, with a minimum age of 24 and a maximum of 42. Most women were between 30 and 33 years old (47.95%) (Table 1). The mean BMI classification indicated that most participants (64.79%) were of normal weight (18.5–24.9 kg/m²), while 29.58% were overweight or obese (BMI ≥ 25) and 5.63% were underweight (BMI < 18.5). Regarding gravidity, 29.17% of women were primigravida while 70.83% were multigravida. Among those with available parity data (N = 66), 34 women (52.3%) were nulliparous, 26 (40%) were primiparous, and 5 (7.7%) were multiparous.

Table 1.

Socio-demographic characteristics of participants (N = 73).

Age ( N = 73)
24–29	21/73 (28.77%)
30–33	35/73 (47.95%)
35+	17/73 (23.29%)
BMI (N = 71)
Underweight (< 18.5)	4 (5.63%)
Normal weight (18.5–24.9)	46 (64.79%)
Overweight/obese (25+)	21 (29.58%)
Gravidity (N = 72)
Primigravida	21 (29.17%)
Multigravida	51 (70.83%)
Parity (N = 65)
Nulliparous	34 (52.3%)
Primiparous	26 (40%)
Multiparous	5 (7.7%)
Women’s level of education (N = 73)		Male spouse level of education (N = 73)
High school level or less	13/72 (16.67%)	High school level or less	17/72 (22.22%)
Bachelor’s degree or its equivalent	30/72 (41.67%)	Bachelor’s degree or its equivalent	33/72 (46.83%)
Master’s degree or its equivalent	29/72 (40.28%)	Master’s degree or its equivalent	22/72 (30.56%)
Doctorate degree or its equivalent	1/72 (1.39%)	Doctorate degree or its equivalent	1/72 (1.39%)
Women’s employment status (N = 73)		Male spouse employment status (N = 73)
Unemployed	17/72 (25%)	Unemployed	3/72 (4.17%)
Employed	54/72 (75%)	Employed	69/72 (95.83%)
Full-time	40/55 (72.73%)	Full-time	46/69 (66.67%)
Part-time	4/55 (5.45%)	Part-time	3/69 (4.35%)
Self-Employed	12/55 (21.82%)	Self-employed	21/69 (28.99%)
Total household monthly income in LBP (N = 72)
Less than $800		2/72 (2.78%)
$800-$1600		3/72 (4.17%)
$1600-$3500		20/72 (27,78%)
$3500 and above		35/72 (48.61%)
Rather not say		12/72 (16.67%)
Main breadwinner for household (N = 72)
Husband		31/72 (43.06%)
Wife		2/72 (2.78%)
50–50		39/72 (54.17%)
MacArthur subjective social status score (N = 72)
Within community		Nationally
Low	-	Low	10 (13.89%)
Medium	37 (51.39%)	Medium	28 (38.89%)
High	33 (45.83%)	High	32 (44.44%)
Rather not say	2 (2.78%)	Rather not say	2 (2.78%)

In terms of education level, 41.67% of women held a bachelor’s degree, 40.28% had a master’s degree, 16.67% had a high school diploma or lower, and 1.39% had a doctorate. Male spouses had similar educational levels, with 46.83% holding a bachelor’s degree, 30.56% having a master’s degree, 22.22% completing high school or lower, and 1.39% holding a doctorate. Concerning employment status, 75% of women were employed, with 72.73% working full-time. In contrast, 95.83% of male spouses were employed, with 66.67% in full-time positions and 28.99% self-employed. Household monthly income data revealed that 48.61% of participants reported earnings above $3,500 per month, while 27.78% earned between $1,600–$3,500, 4.17% between $800–$1,600, and 2.78% less than $800. Additionally, 16.67% of participants preferred not to disclose their income. Regarding financial contribution to the household, 54.17% of couples reported sharing financial responsibilities equally, while 43.06% identified the husband as the primary breadwinner. Social status, assessed using the MacArthur Subjective Social Status Scale, showed that 51.39% of participants rated their status as medium within their community and 45.83% as high. On the other hand, 44.44% perceived their social status compared to the rest of the country as high, 38.89% as medium, and 13.89% as low.

Cord blood heavy metal concentrations

Table 2 presents descriptive statistics for the toxic and essential heavy metals measured in 74 cord blood samples. All metals were detected in all of the samples except for cadmium and antimony, which were detected in 79.9% and 98.6% of samples, respectively. Among toxic metals, mercury (mean = 0.83 µg/L), chromium (mean = 0.79 µg/L) and lead (mean = 0.72 µg/dL) had the highest average concentrations. For essential HMs, zinc (mean = 1948 µg/L), copper (mean = 534 µg/L), and selenium (mean = 114.99 µg/L) were most abundant.

Table 2.

Descriptive statistics of heavy metal concentrations in cord blood (N = 74).

Heavy metal	Limit of detection (LOD)	Limit of quantification (LOQ)	Detection frequency (n,%)	p50	p90	Interquartile range (p25-p75)	Min - max	Mean (SD)
Toxic heavy metals (N = 74)
Antimony (µg/L)	0.01	0.03	73 (98.6%)	0.04	0.07	0.02–0.05	0.005–0.64	0.474 (0.075)
Arsenic (µg/L)	0.007	0.023	74 (100%)	0.237	1.11	0.112–0.477	0.056–4.62	0.486 (0.782)
Chromium (µg/L)	0.09	0.3	74 (100%)	0.59	0.79	0.53–0.65	0.44-15	0.7897 (1.6659)
Nickel (µg/L)	0.04	0.13	74 (100%)	0.55	0.74	0.24–0.55	0.1–12.7	0.6264 (1.4635)
Cadmium (µg/L)	0.007	0.023	59 (79.9%)	0.009	0.16	0.007–0.013	0.0035–0.042	0.1038 (0.0066)
Mercury (µg/L)	0.006	0.02	74 (100%)	0.71	1.55	0.455–1.13	0.155–2.41	0.8343 (0.5233)
Thallium (µg/L)	0.0009	0.003	74 (100%)	0.0082	0.012	0.071 − 0.0106	0.0045–0.0334	0.009272 (0.003694)
Lead (µg/dL)	0.04	0.13	74 (100%)	0.545	1.19	0.392–0.754	0.238–5.44	0.719 (0.73)
Essential heavy metals
Molybdenum (µg/L)	0.02	0.07	74 (100%)	0.65	1	0.53–0.79	0.34–1.35	0.6837 (0.203)
Manganese (µg/L)	0.02	0.07	74 (100%)	31.9	51.3	25.8–41.2	14.6–82.9	34.012 (13.653)
Cobalt (µg/L)	0.005	0.017	74 (100%)	0.122	0.234	0.097–0.169	0.044–0.574	0.1479 (0.0821)
Copper (µg/L)	0.3	1.0	74 (100%)	532	608	497–563	435–697	533.8267 (56.0085)
Zinc (µg/L)	4	13	74 (100%)	1900	2330	1710–2080	1370–2840	1948.4 (306.1668)
Selenium (µg/L)	0.06	0.2	74 (100%)	114	133	106–126	85.8–144	114.99 (13.075)

Spearman’s correlation and heatmap cluster analysis

The non-parametric Spearman correlation matrix (Fig. 1) was used because the data is not normally distributed. With a significance level of p < 0.01, moderate positive correlations were observed between Antimony and Chromium (ρ = 0.36, p = 0.002), Antimony and Nickel (ρ = 0.31, p = 0.007), Antimony and Thallium (ρ = 0.32, p = 0.005), Arsenic and Mercury (ρ = 0.41, p < 0.001), Chromium and Nickel (ρ = 0.36, p = 0.001), Cadmium and Cobalt (ρ = 0.31, p = 0.007), Mercury and Lead (ρ = 0.39, p = 0.001), Zinc and Copper (ρ = 0.4, p < 0.001). At a significance level of p < 0.05, weak positive associations were identified between Chromium and Zinc (ρ = 0.26, p = 0.027), Cadmium and Nickel (ρ = 0.29, p = 0.012), Cadmium and Mercury (ρ = 0.24, p = 0.037), Cadmium and Thallium (ρ = 0.26, p = 0.023), Cobalt and Nickel (ρ = 0.23, p = 0.048), Mercury and Zinc (ρ = 0.28, p = 0.015), Cobalt and Thallium (ρ = 0.27, p = 0.019), Lead and Manganese (ρ = 0.3, p = 0.011), Copper and Manganese (ρ = 0.29, p = 0.013), Zinc and Manganese (ρ = 0.24, p = 0.037), Selenium and Manganese (ρ = 0.29, p = 0.011), Selenium and Copper (ρ = 0.28, p = 0.016). The rest of the associations were not found to be statistically significant.

Fig. 1.

Heatmap and Spearman correlation for the heavy metal concentrations.

Bivariate analysis

Bivariate associations between maternal, environmental, and dietary factors and cord blood heavy metal levels

The bivariate analyses identified a wide range of maternal, environmental, and dietary factors linked to heavy metal concentrations in cord blood samples (Table 3). Several toxic metals were linked to household exposures. For instance, Antimony and Nickel concentrations were significantly associated with types of fuel used for heating (diesel and wood burning, respectively), while Thallium, Lead, and Mercury levels were associated with recent home renovation works. Residential proximity to environmental sources of pollution such as agricultural lands, car parks highways, and factories was also associated with elevated levels of toxic heavy metals, namely Antimony and Arsenic. Among the essential heavy metals, lifestyle factors such as smoking history and household characteristics such as exposure to indoor smoking and recent home renovations were found to be statistically significant factors associated with Manganese, Copper, and Zinc levels in cord blood. Various health-related factors were associated with specific heavy metal concentrations; for instance, Thyroid disease, whether self-reported or in family history, showed statistically significant associations with Chromium and Antimony while allergic conditions such as asthma, pollen allergies, and skin allergies were linked to Nickel, Thallium and Lead cord blood levels. Dietary patterns also emerged as significant factors: intake of fish, nuts, commercial sweets, or butter was associated with levels of Antimony, Nickel, Molybdenum, Cobalt, Copper.

Table 3.

Statistically significant bivariate associations between socio-demographic, environmental, health-related, and dietary factors and cord blood heavy metal concentrations. Statistical significance was defined as p < 0.05.

Heavy metal	Variable (p-value)	Test used	Statistic	p-value
Toxic heavy metals
Antimony	Household heating using diesel	Mann-Whitney U	2.149	p = 0.0317
	Proximity to agricultural lands	Mann-Whitney U	-2.135	p = 0.033
	Proximity to highways	Mann-Whitney U	-2.176	p = 0.0295
	Thyroid disease	Mann-Whitney U	3.228	p = 0.0012
	Famity History (FH) of Thyroid disease	Mann-Whitney U	12.635	p = 0.0055
	Gynecological History of Uterine fibroids	Mann-Whitney U	2.016	p = 0.043
	Medication allergy	Mann-Whitney U	2.4	p = 0.0161
	Servings of fish per week (3 or more vs. less than 3)	Mann-Whitney U	-2.128	p = 0.0334
	Mediterranean Diet Adherence Score (Low vs. Medium)	Mann-Whitney U	-2.142	p = 0.03
Arsenic	BMI	Kruskall Wallis	6.393	0.04
	Gravidity	Kruskall Wallis	7.162	0.007
	Proximity to car parks	Mann-Whitney U	-1.971	0.048
	Heart disease	Mann-Whitney U	-2.028	0.0426
Chromium	Thyroid disease	Mann-Whitney U	4.012	0.0001
	FH of Thyroid disease	Kruskall Wallis	11.23	0.01
Nickel	BMI	Kruskall Wallis	7.576	0.023
	Household heating using wood burning	Mann-Whitney U	2.133	0.0329
	Pollen allergy	Mann-Whitney U	-2.005	0.045
	Fruit units per day (3 or more vs. less than 3)	Mann-Whitney U	2.112	0.035
	Servings of butter per day (less than1 vs. 1 or more)	Mann-Whitney U	-2.034	0.042
Cadmium	Monthly income	Kruskall Wallis	9.866	0.0427
	FH of blood pressure	Mann-Whitney U	2.068	0.0387
	Self-reported diet low in fat	Mann-Whitney U	-2.269	0.02
Mercury	Age of home	Kruskall Wallis	8.462	0.037
	Plumbing renovation works	Mann-Whitney U	2.39	0.0169
Thallium	Window/Glass renovation works	Mann-Whtiney U	-2.813	0.0049
	Pesticide application (yes/no)	Mann-Whitney U	-2.396	0.0166
	Pesticide application frequency	Kruskall Wallis	10.036	0.0066
	Migraine	Mann-Whitney U	-2.281	0.0225
	Skin allergy symptoms	Mann-Whitney U	-1.972	0.048
Lead	Gravidity	Kruskall Wallis	4.153	0.0415
	Plumbing renovation work	Mann-Whitney U	2.054	0.04
	Asthma	Mann-Whitney U	-2.223	0.0262
	Skin allergy symptoms	Mann-Whtiney U	2.067	0.0387
Essential heavy metals
Molybdenum	Servings of nuts per week (3 or more vs. less than 3)	Mann-Whitney U	-2.137	0.03
Manganese	House location	Kruskall Wallis	14.55	0.0421
	Plumbing renovation works	Mann-Whitney U	-1.993	0.0462
	Household heating using electricity	Mann-Whitney U	1.979	0.0478
	Indoor smoking	Mann-Whitney U	2.78	0.0054
Cobalt	Servings of fish per week (3 or more vs. less than 3)	Mann-Whitney U	-2.135	0.033
Copper	Having ever smoked in the past	Mann-Whitney U	-2.212	0.027
	Recent painting renovation works	Mann-Whitney U	1.975	0.0483
	Blood clots	Mann-Whitney U	2.015	0.0439
	FH of epilepsy	Mann-Whitney U	-2.459	0.0139
	Consumption of commercial sweets and pastries (less than 3 vs. 3 or more)	Mann-Whitney U	2.023	0.04
Zinc	Having ever smoked in the past	Mann-Whitney U	2.389	0.0169
	Proximity to diesel generators	Mann-Whitney	2.174	0.0297
	Proximity to factories	Mann-Whitney	-1.977	0.0481
Selenium	Type of household	Kruskall Wallis	7.648	0.0057
	Insect allergy	Mann-Whitney U	-2.006	0.0449
	Preference for chicken (yes/no)	Mann-Whitney U	2.09	0.03

Multivariate analysis

Toxic heavy metals

The main outputs of the multivariable linear regression (MLR) models for toxic heavy metals are shown in Table 4. Antimony levels were significantly predicted by kidney disease (β = 0.457), explaining 90% of the variance in log-transformed concentrations (R² = 0.906). For Arsenic, positively associated factors included residential proximity to a large parking lot (β = 0.312), green space surrounding the house (β = 0.191), heart disease (β = 0.579), and use of solar panel (β = 0.546), with the model explaining 29.4% of the variance (R² = 0.294). In the Chromium model (R² = 0.315), fruit consumption (β = 0.240), heartburn/reflux (β = 0.310), and family history of high cholesterol (β = − 0.149) were retained as statistically significant factors. The model for nickel (R² = 0.423) revealed high blood pressure (β = 0.446), migraine (β = 0.123), and residential proximity to agricultural lands (β = 0.118) as positively associated factors, while butter consumption (β = − 0.116) and BMI (β = − 0.120) were negatively associated. Cadmium levels were positively associated with family history of allergic rash (β = 0.0047) and migraine (β = 0.0050), and negatively with high-fat diet (β = − 0.0066), explaining 28.8% of the variance (R² =0.286). Mercury was significantly associated with family history of kidney disease (β = 0.313), family history of allergic rash (β = 0.206), and higher rate of daily olive oil use (β = 0.271) (R² = 0.315). For thallium, self-reported nutrition quality during pregnancy (β = 0.001), commercial sweets consumption (β = 0.002), and heartburn/reflux (non-significant, retained for relevance) explained 15.1% of the variance. Lastly, Lead concentrations were associated with asthma (β = 0.726), older buildings (β = 0.142), osteoporosis (β = 0.933), and marginally with gravidity (β = − 0.218), with an overall R² of 0.286.

Table 4.

Multivariate linear regressions of toxic heavy metals.

Toxic elements	Factors		Coefficient (β)		95% confidence interval		p-value
Antimony	Weekly servings of fish/shellfish		-0.0228		[-0.046, 0.0003]		0.053
	FH kidney disease		-0.0015		[-0.017, 0.014]		0.842
	Kidney disease		0.4569		[-0.046, 0.0003]		< 0.000
	Constant		0.0605		[0.0354, 0.0855]		0.416
	N obs.	61	R² / Adj. R²	0.906 / 0.901	F(3, 57)	183.18	Model p-value	< 0.001
Arsenic	Proximity to a large parking lot		0.312		[0.088, 0.534]		0.007
	Green area surrounding the house		0.191		[0.008, 0.373]		0.041
	Heart disease history		0.579		[0.135, 1.022]		0.011
	Solar power use		0.546		[0.181, 0.911]		0.004
	Constant		0.093		[-0.072, 0.261]		0.264
	N. obs	64	R² / Adj. R²	0.294/0.246	F(4,59)	6.14	Model p-value	0.0003
Chromium	Employment status of the mother		-0.066		[-0.135, 0.002]		0.058
	Daily fruit unit consumption		0.24		[0.062, 0.418]		0.009
	FH high cholesterol		-0.14		[-0.291, -0.006]		0.040
	Heartburn/reflux		0.31		[0.123, 0.496]		0.002
	Indoor smoking exposure		-0.13		[ -0.284, 0.011]		0.069
	Constant		0.489		[0.185, 0.792]		0.002
	N. obs	61	R² / Adj. R²	0.315/0.252	F(5,55)	5.06	Model p-value	0.0007
Nickel	Daily consumption of butter		-0.116		[-0.197, -0.035]		0.006
	High blood pressure		0.446		[0.0312, 0.215]		0.005
	Proximity to agricultural lands		0.118		[0.144, 0.748]		0.011
	Migraine		0.123		[0.0285, 0.2074]		0.01
	BMI category		-0.12		[0.0285, 0.2074]		0.002
	Constant		0.7010205		[0.0285, 0.2074]		< 0.001
	N. obs	54	R² / Adj. R²	0.42/0.36	F(5,48)	7.03	Model p-value	0.0001
Cadmium	Monthly income		-0.0012		[-0.002, 0.0001]		0.076
	Proximity to factories		-0.0051		[-0.002, 0.0001]		0.107
	Self-reported diet high in fat		-0.0065		[-0.013, -0.0001]		0.045
	FH allergic rash		0.0047		[0.0008, 0.0086]		0.017
	FH migraine		0.005		[0.0009, 0.0091]		0.017
	Constant		0.0125816		[0.008, 0.0170]		< 0.001
	N. obs	52	R² / Adj. R²	0.286/0.208	F(5,46)	3.69	Model p-value	0.0069
Mercury	Spouse level of education		0.082		[-0.012, 0.1763]		0.086
	FH kidney disease		0.313		[0.0727, 0.5524]		0.012
	FH allergic rash		0.206		[0.0799, 0.3324]		0.002
	Natural gas heating		0.135		[-0.026, 0.297]		0.099
	Daily olive oil consumption		0.271		[0.054, 0.4888]		0.016
	Constant		-0.106		[-0.5059, 0.2930]		0.594
	N. obs	51	R² / Adj. R²	0.315/0.239	F(5,46)	4.15	Model p-value	0.0035
Thallium	Self-reported nutrition quality during pregnancy		0.0009		[0.0001, 0.0017]		0.020
	FH heartburn reflux		0.001		[-0.0003, 0.0023]		0.141
	Weekly consumption of commercial sweets		0.0019		[0.0002, 0.0035]		0.141
	Constant		0.0031		[-0.0007, 0.0071]		0.114
	N. obs	60	R² / Adj. R²	0.151/0.105	F(3,56)	3.32	Model p-value	0.0262
Lead	Gravidity		-0.2218		[-0.4585, 0.015]		0.066
	Asthma		0.7265		[0.2341, 1.2189]		0.005
	Age of building		0.1402		[0.033, 0.2474]		0.011
	Osteoporosis		0.9234		[0.0525, 1.794]		0.038
	Constant		1.8661		[1.3786, 2.3536]		< 0.001
	N. obs	59	R² / Adj. R²	0.286/0.233	F(4,54)	5.43	Model p-value	0.001

Essential heavy metals

For the essential heavy metals (Table 5), MLR models revealed several dietary and health-related associated factors of essential heavy metal levels in cord blood. For Molybdenum, higher concentrations were associated with a history of diabetes (β = 0.195) and asthma (β = 0.143) whereas lower levels were linked to increased nuts consumption (β = − 0.079) and olive oil intake (β = − 0.073) as well as a history of kidney disease (β = − 0.237). This model explained approximately 29.3% of the variance in molybdenum levels (R² = 0.293). Manganese levels were inversely associated with the use of electric heating (β = − 0.29, p = 0.006) and frequent sauce consumption (β = − 0.21, p = 0.037), accounting for 19% of the variance (R² = 0.19). Cobalt levels were positively associated with sugary drinks intake (β = 0.039, p = 0.013); the model explained 12% of the variance (R² = 0.12). For the Copper model (R² = 0.398), higher levels were found among participants with greater gravidity (β = 0.079), higher consumption of commercial sweets (β = 0.089) and higher sugary drinks intake (β = 0.132) whereas lower concentrations were observed with recent painting renovation works exposure (β = − 0.078) and higher parity (β = − 0.095). The zinc model (R² = 0.254) showed positive associations with daily vegetable intake (β = 0.088), and inverse associations with poor self-reported nutritional quality during pregnancy (β = − 0.046), self-reported fat-rich diets (β = − 0.137), and having a parking that is adjacent to one’s house (β = − 0.105). Finally, selenium concentrations were higher among those reporting a family history of heartburn or reflux (β = 0.099), personal history of asthma (β = 0.134), and frequent commercial sweets intake (β = 0.098, while a preference for chicken, turkey or rabbit over veal, pork, hamburger or sausage was inversely associated; the Selenium model explained 32.7% of the variance (R² = 0.327).

Table 5.

Multivariate linear regressions of essential heavy metals.

Essential elements	Factors		Coefficient (β)		95% confidence interval		p-value
Molybdenum	Weekly consumption of nuts		-0.079		[-0.1369, -0.0221]		0.007
	Use olive oil as main fat		-0.072		[-0.1298, -0.0155]		0.014
	Diabetes		0.195		[0.0387, 0.3521]		0.015
	Asthma		0.143		[0.0159, 0.2719]		0.028
	Kidney disease		-0.237		[-0.4542, -0.0204]		0.033
	constant		0.000		[0.6058, 0.8462]		< 0.001
	Nb Obs.	60	R² / Adj. R²	0.293/0.228	F(5,54)	4.49	Model p-value	0.0017
Manganese	Electric heating		-0.287		[-0.4885, -0.0850]		0.006
	Sofrito seasoning per week		-0.205		[-0.3971, -0.0133]		0.037
	Constant		4.02		[3.6552, 4.4022]		<0.001
	Nb obs.	53	R² / Adj. R²	0.189/0.157	F(2,50)	53	Model p-value	0.0052
Cobalt	Cup of vegetables per day		0.039		[0.0083, 0.0691]		0.098
	Sugary/carbonated drinks per day		0.039		[-0.0073, 0.0854]		0.013
	Constant		0.028		[-0.0520, 0.1083]		0.485
	Nb Obs.	60	R² / Adj. R²	0.11/0.088	F(2,57)	3.88	Model p-value	0.0263
Copper	Recent painting works		-0.078		[-0.1359, -0.0207]		0.009
	Gravidity		0.079		[0.0065,0.1521]		0.033
	Parity		-0.094		[-0.1508, -0.0391]		0.001
	Commercial sweets consumption per week		0.089		[0.0187, 0.1602]		0.014
	Sugary/carbonated drinks per day		0.132		[0.0396, 0.2253]		0.006
	Constant		6.05		[5.8556, 6.2562]		< 0.001
	Nb obs.	48	R² / Adj. R²	0.398/0/326	F(5,42)	5.55	Model p-value	0.0006
Zinc	Sugary/carbonated drinks per day		0.107		{-0.0076, 0.2223]		0.066
	Cup of vegetables per day		0.088		[0.0085, 0.1677]		0.031
	Self-reported diet high fat		-0.136		[-0.2841, 0.010]		0.067
	Nutrition during pregnancy		-0.046		[-0.0869, -0.0046]		0.030
	Presence of residential parking		-0.105		[-0.1826, -0.0271]		0.009
	Constant		7.63		[7.3601, 7.8999]		< 0.001
	Nb obs	50	R² / Adj. R²	0.253/0.168	F(5,44)	2.99	Model p-value	0.02
Selenium	Commercial sweets consumption per week		0.097		[0.0281, 0.1668]		0.002
	Chicken/turkey preference		-0.08		[-0.1345, -0.0259]		0.005
	Asthma		0.134		[0.0074, 0.2605]		0.038
	FH heartburn		0.099		[0.0391, 0.1581]		0.002
	Constant		4.72		[4.6051, 4.8423]		< 0.001
	Nb. Obs	61	R² / Adj. R²	0.327/0.278	F(4,56)	6.8	Model p-value	0.0002

Discussion

This study provides the first comprehensive assessment of multiple toxic and essential heavy metals in cord blood samples from Lebanese newborns and identifies several sociodemographic, health-related, and dietary factors of fetal exposure. All tested metals were detected in every blood sample, except Antimony and Cadmium, which were still present in the vast majority of participants (98.6% and 79.9%, respectively). Among toxic heavy metals, Mercury (median = 0.71 µg/L), Chromium (median = 0.59 µg/L), Nickel (median = 0.55 µg/L) and lead (median = 0.545 µg/dL) exhibited the highest average concentration, while Zinc (median = 1900 µg/L), Copper (median = 608 µg/L), and Selenium (mean = 114 µg/L) dominated among essential elements. Because cord blood reflects fetal circulation at birth, the concentrations observed for each metal are interpreted as indicators of in-utero exposure during sensitive development windows rather than maternal body burden alone. Heavy metals can cross the placental barrier with relative ease and can induce oxidative stress, disturbing cellular homeostasis, compete with essential proteins for binding sites, leading to the impairment of several biological processes²³. Moreover, adverse pregnancy outcomes, such as low birth weight, have been strongly associated with exposure to toxic heavy metals, namely Arsenic, Cadmium, and Lead^24–26 Additionally, several heavy metals have been linked to neurodevelopmental, cognitive and behavioral issues in children, such as delayed learning and memory deficits^27–30. These health effects can have severe and lifelong health consequences, underscoring the urgency of identifying modifiable factors of exposure to inform targeted public health interventions, especially in the Eastern Mediterranean region, where evidence on this issue remains scarce.

Toxic heavy metals

Antimony

In our study, the median concentration of Antimony was 0.04 µg/L, which is lower than the median concentrations found in Serbia (0.75 µg/L), Japan (0.93 µg/L), China (4.97 µg/L), and Spain (11.22 µg/L)^30–33. In the multivariate model, maternal kidney disease was the strongest predictor of higher Sb levels, consistent with the metal’s nephrotoxic profile and reduced renal clearance³⁴. Despite relatively low doses, the identification of Sb is concerning because it has been linked in the literature to reproductive toxicity and genotoxicity^35,36.

Arsenic

Arsenic concentrations were found at 0.237 µg/L, which is lower than concentrations found in analyses done in Flanders, South Africa, and Spain (0.54 µg/L, 0.46 µg/L, and 1.2 µg/L, respectively)^4,37,38. Residential proximity to large parking lots was positively associated with arsenic levels, suggesting traffic-related emissions as a relevant source, consistent with evidence from the European Commission’s Working Group on Arsenic, Cadmium, and Nickel compounds, which shows higher arsenic concentrations in urban, high-traffic areas compared to rural settings³⁹. Moreover, residential greenness was positively associated with arsenic. Although green spaces can reduce particulate pollution^40,41, this finding could indicate a contaminated soil-related or water-related exposure to Arsenic. The association between arsenic and maternal heart disease aligns with evidence linking chronic exposure to low levels of arsenic with cardiovascular health outcomes, as shown in a large Bangladeshi cohort (n = 11,000 adults)⁴². Lastly, the positive association of solar panel usage and arsenic levels is noteworthy but should be looked at cautiously, as solar panels themselves are not typically a source of arsenic. However, having solar panels might be correlated with other unmeasured household or behavioral factors.

Chromium

The median concentration of chromium in our cohort was 0.59 µg/L, which is higher than the concentrations recorded in Spain (0.2 µg/L)⁴ but lower than levels measured in Poland (8.45 µg/L)⁴³, in Spain (0.99 µg/L)³⁰, and in China (3.42 µg/L)³¹. Chromium levels in cord were associated with the mother’s daily fruit consumption; this aligns with findings from Bocca et al. (2020)¹⁰, which found fruit consumption to be associated with higher Chromium urinary excretion. Interestingly, we found that mothers with higher chromium levels were more likely to suffer from heartburn/reflux symptoms; this finding warrants further investigation because even though there is evidence from human and animal studies that suggests that Chromium may increase the risk of gastrointestinal malignancies and non-malignant conditions such as colonic inflammation and intestinal damage^44–46, there is no evidence directly linking chromium exposure to heartburn symptoms. Moreover, we found that individuals with a family history of high cholesterol had lower chromium levels; however, this association may also be confounded by dietary restrictions, increased health consciousness or other unmeasured confounding variables.

Nickel

The median Nickel concentration (0.55 µg/L) was lower than the concentrations recorded in Spain (0.61 µg/L and 1.2 µg/L)^4,30 and China (1.4 µg/L)⁴⁷. In the multivariate model, residential proximity to agricultural lands was associated with higher Ni levels, suggesting exposure through pesticide use, soil content, and/or dust inhalation in semi-rural farming areas, consistent with concerns about Ni contamination of agricultural ecosystems and food systems⁴⁸. Nickel was also positively associated with high blood pressure, which aligns with evidence linking Ni exposure to oxidative stress and cardiovascular effects, although causality remains uncertain⁴⁹. On the other hand, the association between migraines and higher nickel levels supports growing evidence connecting heavy metals to neurological symptoms. For instance, a case-control study in 2024 by Vural et al. compared the heavy metal and trace element profile in patients with migraine and healthy individuals, and they found Nickel levels (alongside Arsenic, Cobalt, and Lead) to be higher in patients with migraines⁵⁰. Furthermore, lower Ni levels among women with higher BMI replicate patterns observed in a previous biomonitoring study¹⁰. Interestingly, butter consumption was inversely associated with Nickel cord blood levels, indicating potential broader dietary patterns differences that are influencing exposures to heavy metals.

Cadmium

Cadmium had the lowest detection rate of all tested heavy metals (79.9%), with a median concentration was 0.009 µg/L, comparable to Spain (0.01 µg/L)³⁰, China (0.15 µg/L)⁵¹, and South Africa (0.02 µg/L)³⁸ but lower than levels reported in Saudi Arabia (0.704 µg/L)⁵² and another study in Spain (0.5 µg/L)¹⁰. The relatively low levels observed in cord blood may partly reflect the placenta’s role as an effective barrier against cadmium transfer, as placental metallothionein can retain cadmium and limit its passage to the fetal circulation^53,54.

In the multivariate model, a high-fat diet was inversely associated with cadmium levels, likely reflecting reduced intake of vegetables and cereals (major dietary sources of Cd)⁵⁵. Additionally, positive associations with family history of allergic rash and migraine align with evidence on Cadmium’s immunotoxicity and neurotoxicity. For example, the MOCEH study birth cohort in Korea found that higher cord blood Cd is associated with atopic dermatitis in 6-month-old infants⁵⁶. Similarly, another study in 2015 found that Cadmium has a role in generating considerable oxidative stress in patients with acute migraines⁵⁷.

Mercury

Mercury concentrations in our cohort (median: 0.71 µg/L) were comparable to those reported in Spain (0.5–0.67 µg/L)^4,30 and Canada (0.8 µg/L)⁵⁸, but lower than levels observed in Saudi Arabia (2.8 µg/L)⁵², South Korea (5.58 µg/L)⁵⁹, and Spain (7.66 µg/L)⁶⁰. In multivariate models, family history of kidney disease and allergic rash were positively associated with mercury levels, consistent with mercury’s documented nephrotoxic and immunomodulatory effects^61,62. As with cadmium, these familial associations may reflect shared environmental exposures or susceptibility. Olive oil consumption was also associated with higher mercury levels, echoing findings from a recent study linking olive oil intake with mercury biomarkers⁶³, likely reflecting broader adherence to a Mediterranean diet that includes fish and seafood.

Thallium

The median concentration of Thallium in our cord blood samples was 0.008 µg/L, lower than concentrations detected in Italy (0.02 µg/L)⁶⁴, Eastern Croatia (0.05 µg/L)⁶⁵, and China (0.0384 µg/L)⁶⁶. In the Thallium model, a positive association was identified between Thallium cord blood levels and self-reported quality of nutrition during pregnancy; this aligns with previous studies that suggest that well-intentioned healthy eating habits may still pose a risk of exposure to unrecognized toxicants. In the ELFE study, mothers who had “healthier diets” actually had higher exposure to heavy metals⁶⁷, and this is primarily driven by higher intakes of fruits and vegetables⁶⁸; rice and whole grains⁶⁹, and tea leaves⁷⁰.

Lead

In our study, the median lead concentration was 0.545 µg/dL, similar to levels reported in Spain (0.79 µg/dL)⁴ and Iran (0.65 µg/dL)⁷¹ but lower than concentrations documented in China (3.23 µg/dL)⁵¹, Poland (2.56 µg/dL)⁴³, and Saudi Arabia (2.057 µg/dL)⁵². Maternal pre-pregnancy asthma was positively associated with cord blood lead, consistent with evidence showing that asthmatic individuals are more likely to present elevated blood lead levels⁷². Moreover, the association with osteoporosis reflects known lead biokinetics, as lead stored in bone can be mobilized during periods of bone demineralization, including pregnancy. A study in 2019 by Wang et al. (2019) found that higher lead exposure is linked to reduced bone mineral density and increased fractured risk⁷³. Finally, older housing was associated with higher lead levels, aligning with literature showing that deteriorating lead-based paint, old plumbing, and contaminated dust contribute to residential exposure, particularly in homes built before regulatory changes⁷⁴.

Essential heavy metals

Molybdenum

Limited studies have measured Molybdenum in cord blood; however, in our cohort, it had a median concentration of 0.65 µg/L, similar to Austria (0.6 µg/L)⁷⁵ but lower than France (1.44 µg/L)⁷⁶, in China (1.06 µg/L)⁷⁷, and in Serbia (1.31 µg/L)³³. In multivariate models, nut consumption was inversely associated with Mo levels, likely due to the high phytic acid content of nuts, which can chelate minerals and reduce absorption^78,79. Lower Mo levels among women consuming more olive oil are consistent with findings from Notario-Barandiaran et al. (2024), who observed higher urinary excretion of several metals, including Mo, among individuals adhering to a Mediterranean-style diet⁶³, suggesting increased elimination rather than decreased intake. Higher Mo concentrations in mothers with diabetes or asthma align with studies linking Mo-dependent enzymes such as the molybdenum cofactor and xanthine oxidoreductase to oxidative stress processes relevant to these conditions^80,81. Finally, the inverse association with kidney disease echoes evidence that Mo’s antioxidant properties may influence renal handling and reduce circulating levels in individuals with impaired kidney function⁸².

Manganese

The median concentration of Manganese in our study was 31.9 µg/L. Similar concentrations were reported in Canada (31.9 µg/L)⁵⁸, in Poland (31.43 µg/L)⁴³, in Flanders (31.2 µg/L)³⁷, South Africa (34.9 µg/L)³⁸, and in Germany (28.8 µg/L)⁸³. The inverse association between electric heating and manganese levels may reflect reduced indoor air contamination from combustion-related forms of heating, such as wood or fuel-based heating, which are known to contribute to metal exposure indoors. As for the inverse association between Sofrito consumption and Manganese levels in cord blood, it might be explained by the high polyphenol content of tomato-based sauces, which might chelate metals^84,85.

Cobalt

In the current study, the median concentration of Cobalt was 0.122 µg/L, which is approximately half of the concentrations recorded by in South Africa (0.27 µg/L)³⁸, in Spain (0.3 µg/L), and China (0.30 µg/L)⁷⁷. The multivariate model for Cadmium, when controlling vegetable intake, showed a positive association between sugary drinks consumption and Cobalt levels in cord blood; this may be explained by the fortification of certain beverages, such as vitamin-enhanced waters and plant-based milks with B12 (Cobalamin), which contains Cobalt. Moreover, small traces of Cobalt (< 0.03 µg/L) might also leach into hot beverages when stored in plastic or paper containers⁸⁶.

Copper

The median Copper concentration in our cohort (532 µg/L) was higher than levels reported in Spain (367.77 µg/L)³⁰, China (300 µg/L)⁷⁷, and Serbia (274 µg/L)³³. In multivariate models, consumption of commercial sweets and pastries was positively associated with cord blood Copper, consistent with evidence showing that wheat flour, which is common in these products, contains relatively high copper levels⁸⁷. Sugary and carbonated drink consumption also predicted higher Copper levels, although this contrasts with a short-term intervention showing reduced serum copper following intake of sweetened beverages⁸⁸; differences in study duration, populations, and pregnancy-related copper metabolism may explain this discrepancy. Higher gravidity was associated with higher Copper concentrations, whereas greater parity predicted lower levels, suggesting pregnancy-related changes in copper storage and mobilization. Finally, recent home painting was inversely associated with Copper, potentially reflecting increased exposure to lead in older paint, as lead is known to enhance urinary copper loss and impair copper-dependent enzymes⁸⁹.

Zinc

The median Zinc concentration in our cohort (1900 µg/L) was similar to levels reported in Poland (1812 µg/L)⁴³ and lower than those documented in Spain (2311 µg/L)⁴ and South Africa (2548 µg/L)³⁸, and higher than levels documented in Germany (1340 µg/L)⁸³. Higher vegetable consumption was associated with increased Zinc levels, consistent with evidence showing that leafy vegetables, legumes, fruits, and whole grains are important dietary sources of zinc⁹⁰. Women who reported “poor” nutritional quality during pregnancy also had higher Zinc levels; this may reflect greater intake of meat and processed meats, which contain highly bioavailable zinc⁹¹, while phytate-rich “healthier” foods such as cereals, corn, and rice can hinder zinc absorption⁹². Finally, lower Zinc levels in households with attached garages may indicate interactions with traffic-related pollutants or oxidative stress from co-exposure to other metals that can suppress zinc uptake⁹³.

Selenium

The median Selenium concentration in our cohort (114 µg/L) was similar to levels reported in Spain (100 µg/L)⁴ and South Africa (111 µg/L)³⁸, and lower than those found in Poland (156 µg/L)⁴³. Commercial sweets and pastries were positively associated with Selenium levels, likely due to the selenium content of wheat flour, which efficiently accumulates selenium from soil⁹⁴; this is consistent with national soil data showing medium-to-high selenium levels in Lebanon⁹⁵. This association may also reflect broader dietary patterns or consumption of fortified processed foods, as previously observed among lower socioeconomic groups with limited access to meat or seafood¹⁰. Conversely, women who preferred red meat products (e.g., veal, pork, hamburger, sausage) had higher Selenium levels than those favoring poultry, consistent with the higher selenium content of these meats^96,97. Finally, positive associations with maternal asthma and a family history of heartburn may relate to selenium’s roles in antioxidant and immune pathways, though the mechanisms remain unclear and may involve unmeasured confounding.

This study has several limitations. Although the EELI cohort is longitudinal, this analysis relied on a single measurement at birth using one biological matrix (cord blood), limiting the ability to capture temporal variability during pregnancy. The sample size was modest and not fully representative of Lebanon, which may reduce generalizability and limit the detection of subtler associations. Additionally, many exposure were self-reported through phone-based questionnaires, making the exposure data susceptible to recall and social desirability bias. Given the absence of direct environmental measurements, associations are not interpreted as reflecting single mechanistic pathways but rather cumulative and overlapping exposure environments characteristic of daily life during pregnancy. Despite these limitations, this early-phase pilot study provides essential groundwork for understanding prenatal exposure to heavy metals in Lebanon. It is among the first to measure a broad panel of toxic and essential metals in cord blood nationally, offering valuable insight into fetal exposures in a context with limited environmental monitoring. Beyond its geographic and scientific importance, it is also worth mentioning that the study offers a rare window into prenatal exposure to toxic and essential heavy metals in a unique timeframe. Many of the participants were pregnant and gave birth during a time marked with compounded crises: the COVID-19 pandemic, the August 4 Beirut port explosion, and Lebanon’s economic collapse. This makes the data a unique time capsule of fetal exposures during an unprecedented period of acute vulnerability. Importantly, several identified factors are modifiable at the household, environmental, or policy level, highlighting opportunities for targeted interventions to reduce toxic metal exposure while supporting adequate intake of essential elements. Finally, cord blood biomonitoring serves as a powerful risk-communication tool, underscoring the need for stronger protections for pregnant women and their infants.

Methodology

Cohort description

The current study is nested within the Environmental Exposures in Lebanese Infants (EELI) birth cohort, a prospective birth cohort established in 2021 to investigate environmental exposures and health outcomes among pregnant women and their offspring in Lebanon. Participants were recruited during routine antenatal care visits at a partnering obstetrics and gynecology clinic, with enrollment occuring during the second and third trimesters of pregnancy. Detailed descriptions of the cohort design, recruitment strategies, and baseline characteristics have been previously published⁹⁸.

Eligibility criteria include being Lebanese, being 18 years and above, living in the Beirut and Mount Lebanon region for more than 5 years, being able to read, write, and speak Arabic, English, or French, and planning to give birth at the participating hospital, willing to provide informed consent and biological specimen for the EELI study. The Beirut and Mount Lebanon regions were selected due to high population density, heterogenous residential and environmental conditions, and proximity to the partner clinic and hospital.

Exclusion criteria include expecting to move outside of the Beirut or Mount Lebanon region or the country within less than 4 months, spending less than 80% of their nights in their indexed house in the Beirut or Mount Lebanon region, giving birth to infants before 35 weeks of gestation or with major congenital abnormalities or respiratory distress syndrome. Cord blood samples collected at delivery were used to assess prenatal exposure, as they reflect direct fetal exposure at birth. All participants provided written informed consent, and the study protocol was approved by the Ethics Committee of Hotel-Dieu de France University Hospital (Approval number: CEHDF 1683).

Study design

Within the EELI study, demographic, environmental, health, and socioeconomic data were collected using standardized questionnaire. Due to COVID-19 precautions, questionnaires were administered primarily online as computer-assisted telephone interviews, with an option for self-administration when preferred. Biospecimens, including cord blood, venous blood, meconium and colostrum samples, were obtained prenatally and at delivery.

For the present analysis, a subset of (n = 74) pregnant women enrolled between 2021 and 2022 was included. The selection of metals was based on an exploratory biomonitoring framework, including a standardized panel of toxic and essential elements commonly measured in human biomonitoring studies and routinely analyzed by the partner laboratory, with the aim of characterizing prenatal exposure patterns in a data-scarce setting.

Sample collection and analysis

Cord blood samples were collected immediately after delivery using standardized procedures developed prior to study initiation and informed by established international birth cohort protocols. Procedures included consistent collection and handling practices, limited sample manipulation, immediate transport on ice, and storage at − 80 °C until analysis to minimize contamination. As such, samples were drawn into heparinized collection tubes and transported on ice to the research laboratory within 6 h of delivery. Upon arrival, samples were gently homogenized by inversion and aliquoted into pre-labeled cryovials without separation of blood components, then stored at − 80 °C until shipment for analysis at a partner laboratory, RECETOX Laboratories (Masaryk University, Brno, Czech Republic), an internationally recognized analytical facility specializing in human biomonitoring and trace element analysis. All aliquots were prepared using pre-specified volumes to avoid sample waste, and no biological specimens were retained at the analytical laboratory beyond the scope of this analysis.

Sample preparation and laboratory analysis

Concentrations of antimony (Sb), arsenic (As), chromium (Cr), molybdenum (Mo), manganese (Mn), cobalt (Co), nickel (Ni), copper (Cu), zinc (Zn), cadmium (Cd), mercury (Hg), thallium (Tl), lead (Pb), and selenium (Se) in whole cord blood samples were quantified using inductively coupled plasma tandem mass spectrometry (ICP-MS/MS). Prior to analysis, blood samples were diluted tenfold (0.2 ml blood aliquot diluted to 2 ml) with a solution containing deionized water, a non-ionic surfactant Triton X-100 (0.04%), ammonia (1%), butanol (2%), EDTA (0.04%), and the appropriate internal standards for ICP-MS (20 ng/ml of Sc, Ge, In, Lu and Bi). Quantification of trace elements in the diluted samples was performed using an Agilent 8900 ICP-MS/MS instrument (Agilent technologies) equipped with MicroMist concentric nebulizer, Scott double-pass spray chamber, quartz torch with 2.5 mm i.d. injector and nickel sampling/skimmer cones. The ICP-MS operated at 1600 W input power, had a sampling depth of 9 mm, and a carrier gas (argon) flow rate of 1.15 L/min. To suppress spectral interferences, an Octapole reaction system was employed in collision mode with helium flow 9 ml/min.

Quality assurance and quality control

The validation process for the trace element determination method included analyzing samples spiked with a known quantity of the analytes, as well as analyzing certified reference materials (SERO AS, Norway), specifically Seronorm™ Trace Elements Whole Blood Level 1 and Seronorm™ Trace Elements Whole Blood Level 2. Throughout the analytical sequence, blanks (diluted 0.2 ml aliquot of DI water) and certified reference materials were consistently analyzed. Approximately one blank and one reference material were analyzed for every 10–15 samples. Notably, the ICP-MS measurements yielded recoveries typically ranging from 85% to 115% for both the spiked samples and the certified reference materials.

Furthermore, RECETOX’s trace analytical laboratories hold accreditation for the analysis of trace elements in whole blood according to European standards as certified by the Czech Accreditation Institute (ČSN EN ISO/IEC 17025:2018). Quality control is ensured by successful participation in the several interlaboratory comparisons organized by G-EQUAS, INSTAND EQUAS and ICI-EQUAS under HBM4EU project.

Statistical analysis

Descriptive Statistics were calculated for all heavy metal concentrations, including means, standard deviations, medians, 90th percentiles, and interquartile ranges (IQR). Following EPA guidelines on handling non-detects in environmental risk assessment, non-detect values for Antimony and Cadmium were substituted with half the detection limit (LOD/2)⁹⁹. As metal concentration distribution were deviated from normality, non-parametric tests were applied.

Spearman’s correlation coefficients were computed to assess the relationships between metal concentrations, explore co-occurence patterns and potential shared exposure pathways. Associations between categorical variables and heavy metal concentrations were conducted using Mann-Whitney U tests for binary variables and Kruskal-Wallis tests for multi-category variables.

Multivariable linear regression models were conducted using log-transformed metal concentrations as dependent variables to examine associations with selected socio-demographic, environmental, dietary, and health-related factors. Covariates were included based on theoretical relevance and bivariate associations (p < 0.2). Multicollinearity was assessed using Variance Inflation Factors (VIF), and model fit was evaluated using Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC). Statistical analyses were performed using STATA v12 (StataCorp, College Station, TX, USA) with a statistical significance set at p < 0.05.

Author contributions

EW, GA, and MM conceptualized and designed the study. Data collection was carried out by JA, AA, CK, NA, and NY. JK contributed to the laboratory sample analysis and provided technical expertise for the analytical procedures; they also revised and contributed to the Sample Preparation, Laboratory Analysis, and Quality Assurance and Control sections of the manuscript. EW conducted the data analysis and statistical modeling with input from MM and GA. MM and JA contributed to refining the methodological approach. EW and JA drafted the manuscript, while GA and MM reviewed, revised, and edited it. All authors have read and approved the final version of the manuscript. GA and MM have contributed equally to the manuscript as co-last authors.

Funding

This research was made possible through the Health and Environment Response Agency (HERA) through an EU Fund [No 101137317], as part of the consortium IHEN Project-Horizon-HLTH-2023-ENVHLTH-02. This work was also supported from the European Union’s Horizon 2020 research and innovation program under grant agreement No 857,560 (CETOCOEN Excellence). This publication reflects only the author’s view, and the European Commission is not responsible for any use that may be made of the information it contains. Authors also thank the RECETOX Research Infrastructure (No LM2023069) financed by the Ministry of Education, Youth and Sports for supportive background.

Data availability

The data and materials that support the findings of the study are available upon reasonable request from the corresponding author, MM.

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval

Ethical approval for this study was granted by the Hotel-Dieu de France University Hospital’s Ethics Committee (Approval number: CEHDF 1683), in accordance with the World Health Organization (WHO)’s Standards and Operational Guidelines for Ethical Review of Research Involving Human Participants (2011) and the principles outlined in the Declaration of Helsinki. Informed consent was obtained from all participants, who were assured confidentiality, privacy, and anonymity of their information and identity. Participants were informed of their right to withdraw from the study at any stage, and their withdrawal had no impact whatsoever on the services provided to them. Additionally, all aliquots were prepared with the exact volumes needed to ensure no biospecimen waste was left, and the subcontracted laboratory destroyed any additional remaining volumes of the samples in accordance with established protocols. Written informed consent was obtained from the participants for publication of data.