Heavy metals exposure and risk factors for autism spectrum disorder (ASD) among school children in Makassar City, Indonesia

Abstract

This study investigated the associations between heavy metal exposures and autism spectrum disorder (ASD) in school-aged children from Makassar, Indonesia, using an unmatched case-control design with 30 ASD cases and 30 controls without ASD (aged 6–11 years). Hair samples were analyzed by ICP–MS to quantify mercury (Hg), lead (Pb), and cadmium (Cd) levels. Parental questionnaires collected data on potential confounders, including family income, exclusive breastfeeding, family history of ASD, maternal dietary habits, use of skin-whitening creams, and tobacco smoke exposure. Unconditional logistic regression analysis revealed that children with hair Hg levels below 2.0 µg/g had significantly increased odds of ASD (OR = 19.3; 95 % CI: 1.78–533; p = 0.0312), underscoring a strong association between mercury exposure and ASD risk. Conversely, lower hair Pb levels (< 4.5 µg/g) were linked to reduced odds of ASD (OR = 0.164; 95 % CI: 0.0215–0.980; p = 0.0569), suggesting a contradiction potentially mediated by genetic or epigenetic differences in lead metabolism. Additionally, a positive family history of ASD (OR = 24.7; 95 % CI: 2.20–1195; p = 0.0338) emerged as a robust predictor, while maternal use of skin-whitening creams (OR = 0.119; 95 % CI: 0.0121–0.806; p = 0.0409) and the absence of prenatal tobacco smoke exposure (OR = 9.47; 95 % CI: 1.38–193; p = 0.0494) were significantly associated with ASD risk. These findings highlight the multifactorial etiology of ASD and emphasize that both environmental exposures to heavy metals and specific maternal risk factors substantially influence neurodevelopmental outcomes.

Graphical Abstract

Highlights

• •Children with hair mercury levels of <2.0 µg/g are linked to higher ASD odds, emphasizing maternal mercury exposure's critical role.
• •Lower hair lead levels (<4.5 µg/g) were associated with ASD, suggesting potential genetic or epigenetic variations in lead metabolism.
• •A positive family history of ASD strongly predicts ASD risk, highlighting the role of hereditary and genetic factors.
• •Maternal skin-whitening cream use and the absence of prenatal smoke exposure were linked to ASD, showing multifactorial aetiology of ASD.
• •Findings from Makassar, Indonesia, show ASD risk from toxicants and socio-behavioral factors, urging public health action amid industrial growth.

1. Introduction

Autism Spectrum Disorder (ASD) is a complex neurodevelopmental condition characterized by impairments in social interaction, communication, and the presence of restricted, repetitive behaviors [2], [3], [12]. Over the last few decades, the reported prevalence of ASD has increased noticeably, a trend that cannot be fully explained by enhanced clinical awareness and improved diagnostic tools alone [5], [26]. This observation has prompted a significant amount of research looking into possible contributing factors, particularly the interactions between genetic predispositions and environmental exposures [16], [17].

Many scientists now believe that prenatal exposure to various environmental toxicants plays a significant role in increasing the risk of ASD. Heavy metals such as lead (Pb), mercury (Hg), and cadmium (Cd) have gained particular attention because of their known neurotoxic effects and their ability to interfere with brain development. For instance, lead is associated with diminished cognitive performance and disrupted synaptic function [4], [22]. Mercury, in both its methylmercury and inorganic forms, can damage mitochondria, raise oxidative stress levels, and disturb normal neurotransmission processes [15], [17], [34]. Although the role of cadmium remains somewhat inconsistent across studies, some evidence suggests that it may promote neuroinflammation and oxidative imbalances [11], [15], [32].

Children are especially vulnerable to such toxic exposures due to their developing physiological systems, higher absorption rates via the respiratory and gastrointestinal systems, and behaviors like hand-to-mouth activity [4], [22]. Maternal exposures further complicate this picture. For example, pregnant women may encounter hazardous chemicals at work or through daily activities, such as using unregulated cosmetic products containing mercury or being exposed to environmental tobacco smoke. These factors can significantly alter the in utero environment by disrupting neuronal proliferation, migration, and synaptogenesis—processes critical for normal brain development [6], [13], [30].

The situation becomes even more concerning in rapidly industrializing regions like Makassar, Indonesia. Here, industrial expansion, uneven enforcement of environmental regulations, and limited public health resources create complex exposure scenarios. Unregulated industrial emissions, combined with cultural practices such as high-frequency seafood consumption and the use of cosmetic products containing mercury, can elevate the body burden of these toxic metals [1], [29], [35]. In Indonesia, for example, local prevalence data indicate that ASD rates have increased from approximately 1 in 1000 births in the early 2000s to nearly 1 in 250 children by 2015 [27], [9], [23]. This stark increase underscores the urgent need for systematic investigations into environmental contributions to ASD.

Recent evidence highlights that maternal exposure to environmental toxicants—including heavy metals and pesticides—can substantially increase the risk of ASD in offspring. The review demonstrated that prenatal exposure during critical periods of fetal brain development—particularly during the second and third trimesters—can disrupt key neurodevelopmental processes through mechanisms such as oxidative stress, neuroinflammation, inhibition of gamma-aminobutyric acid (GABA) signaling, and acetylcholinesterase (AChE) inhibition [25]. These mechanisms collectively impair brain structures responsible for social behavior and cognition—hallmark features of ASD. Particularly, oxidative damage and neurotransmitter imbalances caused by toxicants like lead (Pb) and organophosphates may have long-lasting effects on cortical development, increasing ASD susceptibility even at low exposure levels that are not neurotoxic to adults. This reinforces the urgent need for targeted maternal environmental health interventions in populations at risk.

Moreover, Maleki et al. (2023) noted that the adverse effects of these exposures might be compounded by genetic and epigenetic factors. Genetic polymorphisms in detoxification enzymes—such as those encoded by delta-aminolevulinic acid dehydratase (ALAD) and glutathione-S-transferases (GSTs)—can influence how efficiently heavy metals are metabolized and cleared from the body [10], [7]. Epigenetic modifications triggered by environmental exposures may further regulate the expression of these enzymes, thereby altering an individual’s vulnerability to heavy metals [20], [18]. These findings help to explain why some epidemiological studies report unexpected patterns—such as lower hair lead levels in some ASD populations—possibly reflecting enhanced lead clearance or redistribution mediated by genetic or epigenetic factors.

Additionally, socioeconomic status (SES) is an important factor that shapes the exposure environment. Factors such as family income and maternal education not only determine the likelihood of exposure to environmental contaminants but also influence access to cleaner, regulated consumer products and safer living conditions [1], [15], [19]. In urban centers like Makassar, lower SES may be associated with greater exposure to industrial pollutants and substandard housing, which together contribute to higher overall toxicant exposure.

Given this complex interplay of environmental, genetic, and socio-behavioral factors, our study is designed to address several of these critical gaps by examining hair lead, mercury, and cadmium levels among children with and without ASD in Makassar, alongside key maternal and demographic characteristics. By employing a case-control approach, it aims to identify the extent to which these metals differentiate ASD cases from controls and to ascertain whether specific maternal behaviours (e.g., usage of whitening creams, tobacco smoke exposure) or familial traits (e.g., a family history of ASD) are correlated with the identified metal burdens.

Our approach is aligned with global research efforts that recognize the multifactorial nature of ASD. The integration of environmental measurements with genetic susceptibility and socioeconomic assessments represents a comprehensive strategy to untangle the complex pathways linking toxicant exposures to neurodevelopment. The hope is that such insights will not only clarify the biological underpinnings of ASD but also guide targeted public health interventions. For example, improved regulation of cosmetic products, educational campaigns about safe seafood consumption, and anti-smoking initiatives could be implemented to reduce toxic exposures. Furthermore, understanding the genetic and epigenetic factors that modulate heavy metal metabolism may eventually lead to personalized screening and intervention strategies for at-risk populations.

2. Materials and methods

This study employed an unmatched case-control design to investigate potential associations between exposure to selected heavy metals—namely lead (Pb), mercury (Hg), and cadmium (Cd)—and ASD among school age children in Makassar, Indonesia. Thirty children clinically diagnosed with ASD and 30 children without ASD (control group), aged between 6 and 11 years were enrolled. Although the two groups were matched by sample size, they were not strictly matched by age, sex, or socioeconomic status (SES). These factors were considered potential confounders and were carefully examined and statistically adjusted in subsequent analyses.

2.1. Case group (Children with ASD)

Thirty children were recruited from five special schools (Sekolah Luar Biasa) and one rehabilitation centre for ASD in Makassar. All participants had been diagnosed with ASD (mild, moderate, or severe) by qualified healthcare professionals.

2.1.1. Inclusion criteria

• •Clinically confirmed ASD diagnosis (mild, moderate, or severe) by a professional.
• •Aged approximately 6–11 years, though minor deviations were permitted due to institutional enrolment ages.
• •Parents or guardians provided written informed consent, following detailed explanations of the study’s objectives and procedures.

2.1.2. Exclusion criteria

• •Co-occurring major psychiatric disorders other than ASD (e.g., severe attention-deficit/hyperactivity disorder, schizophrenia).
• •Chronic or acute medical conditions restricting hair sample collection (e.g., scalp infections or significant dermatological issues).
• •Refusal or withdrawal of consent at any stage of the study.

2.2. Control group (Children without ASD)

An additional 30 children were selected from general elementary schools located in the same geographical area of Makassar. None of these children had a known diagnosis of ASD or other neurodevelopmental disorders.

2.2.1. Inclusion criteria

• •Considered neurotypically developing, with no previously reported history of ASD or other significant developmental disorders.
• •Aged approximately 6–11 years.
• •Parents or guardians granted informed consent after a comprehensive overview of the study’s procedures.

2.2.2. Exclusion criteria

• •Clinical or parent-reported neurodevelopmental or significant psychiatric conditions.
• •Medical issues that precluded safe or valid sampling of hair specimens (e.g., dermatological contraindications).
• •Parent or guardian’s decision to decline participation prior to or during data collection.

2.3. Data collection and assessment of confounding factors

Data on a broad spectrum of potential confounding factors were obtained using a structured paper-based questionnaire administered to the parents or guardians. The questionnaire captured the following variables: demographic characteristics (age, gender, and family income), infant feeding history (history of exclusive breastfeeding), familial risk factors (family history of ASD), maternal dietary habits (frequency of seafood consumption, a key exposure route for mercury; and history of vegetables and/or fruits consumption potentially containing pesticide residues), maternal behaviors (frequency of skin-whitening (lightening) cream use, which may indicate exposure to mercury-containing cosmetics), tobacco smoke exposure(both prenatal and postnatal exposure to tobacco smoke were assessed given their potential role in modulating heavy metal exposures and neurodevelopmental outcomes). These variables were selected based on previous research suggesting their relevance in influencing heavy metal exposure and ASD risk [1], [6], [13], [16], [21], [29], [35].

2.4. Hair sampling and laboratory analysis

For every participant, a lock of hair (approximately 0.5 g) was cut from the occipital region, close to the scalp, using sterilized scissors. This samples were carefully pre-processed and stored in sterile and labelled plastic bags to prevent cross-contamination and transported to an accredited reference laboratory operating under ISO/IEC 17025:2005 standards.

In the laboratory, each hair specimen was digested by adding concentrated nitric acid (HNO₃) and heating the mixture at 95°C until complete dissolution. The solution was then cooled, filtered, and diluted with ultrapure water. Concentrations of Pb, Hg, and Cd were measured by Inductively Coupled Plasma–Mass Spectrometry (ICP–MS). The limits of detection (LOD) for Pb and Hg were in the range of 0.001–0.01 µg/g, and for Cd, approximately 0.0005–0.005 µg/g; corresponding limits of quantification (LOQ) were also established and used in quality control measures.

2.5. Statistical analysis and adjustment for confounding factors

To contextualize hair metal concentrations, cut-off values established in previous studies and validated by local pilot data were employed. Specifically, hair Pb levels were classified as either ≥ 4.5 µg/g or < 4.5 µg/g, while Hg and Cd levels were classified as ≥ 2.0 µg/g or < 2.0 µg/g. These thresholds are applied with caution due to the lack of universal consensus on toxicity levels in hair, as well as variability resulting from dietary habits, occupational exposures, and local regulatory practices.

All data were entered into a secure database and analyzed using R version 4.2.3 [33]. The distribution of heavy metal concentrations was examined using the Shapiro–Wilk test, which indicated non-normal distributions for the metals, leading to the use of non-parametric tests for pairwise comparisons (Mann–Whitney U test) where applicable.

To minimize confounding effects, initial comprehensive bivariate analyses were performed to evaluate the association between each potential confounder and both the outcome (ASD status) and the heavy metal concentrations in hair samples. Specifically, variables such as age, gender, family income, history of exclusive breastfeeding, family history of ASD, maternal frequency of seafood consumption, maternal consumption of vegetables and/or fruits that might be contaminated with pesticides, maternal lightening (whitening) cream use, and both prenatal and postnatal tobacco smoke exposure were examined using appropriate statistical tests (e.g., Chi-square tests for categorical variables, Mann–Whitney U tests and Spearman rank correlations for continuous variables). These analyses allowed us to identify candidate covariates that might influence the association between heavy metal exposures and ASD.

To further refine the selection of variables to be included in the logistic regression model and to address issues of overfitting and multicollinearity, we employed the Boruta algorithm for variable selection [8]. Boruta is a robust, all-relevant feature selection method based on random forests that iteratively compares the importance of real predictors with that of randomized shadow features. This process helps to confirm the significance of each variable, ensuring that only those covariates with a demonstrable impact on the outcome and heavy metal levels are retained.

Subsequently, unconditional logistic regression models were constructed to estimate the odds ratios (OR) and 95 % confidence intervals (CI) for associations between heavy metal exposures and ASD. In these models, the aforementioned covariates were included as adjustment variables. This multivariable adjustment was designed to control for potential confounding by ensuring that the observed associations between heavy metal levels and ASD were not spuriously influenced by differences in age, gender, socioeconomic status, infant feeding, familial ASD predisposition, maternal dietary habits, cosmetic use, and tobacco smoke exposure. The final model was selected based on both exploratory analyses and the Boruta feature-selection algorithm, which confirmed the importance of these covariates in explaining the variation in ASD status.

This methodological approach allowed us to examine the independent effect of heavy metal concentrations on ASD risk while minimizing bias due to confounding. Model diagnostics were performed to assess goodness-of-fit, multicollinearity, and potential interaction effects, thereby ensuring the robustness and validity of our statistical inferences.

2.6. Ethical approval

This study was conducted in strict accordance with the ethical principles outlined in the Declaration of Helsinki. The study protocol was reviewed and approved by the Research Ethics Commission of the Faculty of Public Health, Hasanuddin University (Letter Number 5424/UN4.14.1/TP.01.02/2022). All procedures involving human participants were carried out following these ethical guidelines, ensuring the protection of participant confidentiality and the welfare of the children and their families. Informed written consent was obtained from the parents or guardians of all participants prior to enrollment in the study.

3. Results and discussion

A total of 60 children, comprising 30 children diagnosed with ASD and 30 controls without any reported neurodevelopmental disorders. An independent two-sample t-test as shown in Fig. 1, revealed no statistically significant difference in age between the two groups (mean age [ASD] = 9.07 years vs. [Non-ASD] = 8.67 years; t = 0.798, df = 58, p = 0.428), indicating adequate comparability in this demographic variable.

Fig. 2.

Mean comparison of Heavy Metal concentrations in hair samples for Case (ASD) and Control groups.

Fig. 1.

Distribution of children age based on group.

Bivariate analyses were subsequently conducted using Chi-square tests to assess the relationship between potential confounding factors and ASD status as well as their association with heavy metal concentrations. Table 1 reveals that although the male-to-female ratio differed between groups (19 males/11 females in the ASD group vs. 12/18 in controls), this difference was not statistically significant (χ² = 2.403, p = 0.121). A significant difference was observed regarding family income (<7 million vs. >7 million Indonesian Rupiah), with lower income families being more prevalent in the ASD group (χ² = 7.547, p = 0.006). In addition, trend toward lower exclusive breastfeeding in the ASD group was noted (p = 0.060). A significantly higher proportion of ASD cases reported a positive family history (13 vs. 1; χ² = 11.273, p = 0.0008). Moreover, Maternal lightening (whitening) cream use (χ² = 8.208, p = 0.0042) and prenatal tobacco smoke exposure (χ² = 4.389, p = 0.036) were also significantly associated with ASD status. In contrast, maternal frequency of seafood consumption and postnatal tobacco exposure showed trends toward significance (p = 0.055 and 0.068, respectively), while maternal consumption of pesticide-contaminated vegetables/fruits did not differ between groups (p = 1.000).

Table 1.

Chi-square test results for key maternal and environmental exposure variables by ASD status.

Variable	ASD	Non-ASD	χ²	p-value
Gender
Male Female	19 11	12 18	2.4027	0.1211
Family income (IDR)
< 7 billion > 7 billion	18 12	28 2	7.5466	0.006012
Exclusive breastfeeding
No Yes	10 20	3 27	3.5352	0.06008
Family history of ASD
Yes No	13 17	1 29	11.273	0.0007863
Maternal Frequency of Seafood consumption
Frequently (>3 Times/week) No Frequently (<3 Times/week)	24 6	16 14	3.675	0.05523
Maternal consumption of pesticide-contaminated vegetables/fruits
Yes No	26 4	27 3	0	1
Maternal History of Lightening Cream Use
Yes No	14 16	3 27	8.2079	0.004171
Pre-natal exposure to tobacco smoke
Exposed Not exposed	22 8	13 17	4.3886	0.03618
Post-natal exposure to tobacco smoke
Exposed Not exposed	21 9	13 17	3.3258	0.0682

3.1. Heavy Metal Concentrations

Hair samples were analyzed for mercury (Hg), lead (Pb), and cadmium (Cd) concentrations using ICP–MS. As detailed in Table 2, the median (interquartile range) hair Hg level in the ASD group was 1.68 µg/g (1.01–2.46) compared to 0.95 µg/g (0.50–1.60) in controls (Mann–Whitney U, p = 0.03). This elevation in hair Hg aligns with previous findings that neurotoxic metals can be more pronounced in populations experiencing neurodevelopmental challenges [15], [17], [34]. One explanation is that mothers of ASD children may have encountered greater mercury exposure during pregnancy via cosmetic products or dietary intake of methylmercury-containing fish [1], [29]. Although direct causality cannot be inferred from these cross-sectional data, the elevated Hg levels in cases align with prior evidence in both developing and developed settings [14], [28].

Table 2.

Median and Interquartile Range (IQR) of Heavy Metal Concentrations (µg/g) in Hair Samples for ASD and Control Groups.

Heavy Metal	ASD (n = 30) Median (IQR)	Control (n = 30) Median (IQR)	p-value (Mann–Whitney U)
Hg	1.68 (1.01–2.46)	0.95 (0.50–1.60)	0.03
Pb	3.92 (2.15–6.72)	8.57 (4.10–12.33)	0.01
Cd	0.32 (0.12–0.64)	0.28 (0.10–0.58)	0.44

Conversely, hair Pb levels were significantly lower in children with ASD (median 3.92 µg/g, IQR: 2.15–6.72) than in controls (median 8.57 µg/g, IQR: 4.10–12.33; p = 0.01), diverging from the mainstream understanding that lead toxicity often correlates with neurodevelopmental impairments [4], [22]. Such a paradox might result from complex local factors, including possible differences in housing conditions or protective behaviours among families of ASD children [30], [31]. Another hypothesis is that genetic or epigenetic factors relevant to Pb metabolism could facilitate lower hair lead content in ASD subgroups, even if actual environmental contact does not differ substantially [16], [24], [30]. Regardless, the finding reveals that universal assumptions regarding Pb and ASD risk may not always apply in every geographic or cultural context. No significant difference was observed for Cd levels between the two groups (p = 0.44). Cadmium did not show a statistically significant disparity between ASD and control groups, suggesting that in this Makassar population, Cd exposure may be comparatively lower or less influential. Some previous studies have proposed a connection between cadmium and neuroinflammatory or oxidative mechanisms [15], [32], yet this sample did not reflect such an association. Consequently, the data highlight the variability in how different heavy metals manifest in child populations, especially those in rapidly urbanizing areas [19].

Pairwise comparison and correlation plot for three heavy metals—mercury (Hg), lead (Pb), and cadmium (Cd)—in two groups: “Case” (red) and “Control” (teal) as shown in Fig. 3 also confirm the findings. The case group generally displays elevated mercury levels, lower lead levels, and no major difference in cadmium compared to the control group, but importantly shows a strong Pb–Cd correlation. The control group, by contrast, has higher lead on average but a much weaker Pb–Cd association. These patterns could reflect distinct environmental exposures or metabolic handling of metals in children with ASD (case group) versus controls.

Fig. 3.

Pairwise comparison and correlation plot for three heavy metals in Case (ASD) and Control Groups.

Hg and Pb overall correlation is 0.094, meaning there is little to no linear relationship when combining both groups. However, the case group (0.297) shows a mild positive correlation (higher Hg values tend to accompany higher Pb values), whereas the control group (–0.116) exhibits a small negative correlation. Hg and Cd overall correlation is –0.009, effectively near zero. Both case (0.034) and control (–0.011) groups also lack any notable linear relationship between mercury and cadmium. Pb and Cd overall correlation is 0.385, which is moderately positive. Notably, in the case group (0.581), Pb and Cd are strongly correlated, but in the control group (0.103), that relationship is weak. This suggests that in the case group, children who have higher lead levels also tend to have higher cadmium levels, whereas in the control group the two metals show little to no association.

3.2. Association analyses and interpretation

Spearman Rank correlation tests (Fig. 4) conducted within each group showed no noteworthy correlations among Hg, Pb, and Cd in the ASD group. In the control group, a moderate correlation was observed only between Pb and Cd (rho = 0.42, p = 0.02). A separate correlation analysis focusing on maternal variables and child metal levels found that maternal whitening cream use correlated moderately with child hair Hg in the ASD group, offering a plausible route for mercury exposure [1], [29]. Prenatal tobacco smoke exposure also approached significance in correlating with child Hg levels, hinting at combined toxicological stresses on neurodevelopment [6], [13].

Fig. 4.

Heatmap of Spearman Rank Correlation Matrix.

To identify the most relevant predictors for inclusion in the logistic regression model, the Boruta algorithm was applied [8]. This feature selection approach confirmed the importance of variables such as dichotomized hair Hg (≥ 2.0 µg/g vs. < 2.0 µg/g), dichotomized Pb (≥ 4.5 µg/g vs. < 4.5 µg/g), exclusive breastfeeding (ASI_EKS), family history of ASD (RG), family income (X1_n), maternal lightening cream use (X10), and prenatal tobacco smoke exposure (X11). Variables like Cd1, gender, and age group were rejected or deemed tentative.

An unconditional logistic regression model was subsequently fitted with ASD status as the outcome and the confirmed predictors as covariates. As shown in Table 3, it reveals that children with Hg levels below the 2.0 µg/g threshold had an odds ratio (OR) of 19.3 (95 % CI: 1.78–533; p = 0.0312), suggesting a strong association between elevated mercury exposure and ASD risk. Conversely, lower Pb levels (< 4.5 µg/g) were associated with ASD (OR = 0.164, 95 % CI: 0.0215–0.980; p = 0.0569), although this result approached marginal significance. While discordant with many established lead toxicity findings [4], [22], [35], this result may reflect a range of protective or confounding factors unique to Makassar or to families of ASD children [30], [31].

Table 3.

Unconditional Logistic Regression Model Predicting ASD Status.

Variable	OR	95 % CI	p-value
Hair Hg (≥ 2.0 µg/g vs. < 2.0 µg/g)	19.3	1.78 – 533	0.0312
Hair Pb (≥ 4.5 µg/g vs. < 4.5 µg/g)	0.16	0.02 – 0.98	0.0569
Family income (> 7million vs < 7million)	0.11	0.006 – 0.99	0.0736
Family history of ASD (Yes vs. No)	24.7	2.20 – 1195	0.0338
Maternal history of lightening cream use (Yes vs. No)	0.12	0.01 – 0.806	0.0409
Prenatal tobacco smoke exposure (Exposed vs. Not exposed)	9.47	1.38 – 193	0.0494

In the same model, exclusive breastfeeding demonstrated a protective effect on neurodevelopment. Moreover, family history of ASD remained a robust predictor (OR = 24.7, 95 % CI: 2.20–1195; p = 0.0338). This finding aligns with extensive literature emphasizing genetic heritability in ASD, including research on potential gene–environment interactions [28,52]. Maternal lightening cream use was significantly associated with ASD (OR = 0.119, 95 % CI: 0.0121–0.806; p = 0.0409). Prenatal tobacco smoke exposure likewise emerged as a critical predictor, consistent with broader evidence that maternal smoking can exacerbate fetal vulnerability to neurotoxicants, including heavy metals found in tobacco [6], [21].

These results indicate that after adjusting for key confounding factors, both heavy metal exposures and certain maternal/familial characteristics are significantly associated with ASD status in this population. This study provides evidence that environmental exposures to heavy metals, particularly mercury, are associated with an increased risk of ASD in children from Makassar, Indonesia.

3.3. Mercury exposure and ASD

Elevated hair mercury among children with ASD is consistent with prior studies that implicate mercury as a neurotoxicant capable of disrupting mitochondrial function, inducing oxidative stress, and interfering with neurotransmission [15], [17], [34]. The significantly greater odds of ASD associated with mercury concentrations above the 2.0 µg/g threshold underscore the potential role of maternal exposures, possibly via the use of mercury-containing cosmetic products and dietary habits involving frequent seafood consumption. These findings support the need for public health interventions focused on mitigating mercury exposure in vulnerable populations.

In contrast to the established literature associating lead exposure with neurodevelopmental impairments [4], [22], our analysis revealed that lower hair lead levels were found in the ASD group. One plausible explanation is that genetic or epigenetic factors influencing lead metabolism may differ in this population. Variations in genes involved in detoxification pathways, such as those encoding glutathione-S-transferases, may lead to differential accumulation or excretion of lead [16], [30], [31]. Furthermore, protective cultural practices or differences in environmental lead sources, such as variations in housing materials or localized industrial emissions, might contribute to these unexpected findings. It is also possible that the biomarkers used (i.e., hair lead concentration) may not fully capture the complexity of lead exposure dynamics in this setting.

Several biological mechanisms have been proposed to explain the neurotoxic effects of heavy metals, which may contribute to the development of ASD. In our study, we observed an unexpected inverse relationship for Pb levels between children with ASD and controls, prompting further exploration into the underlying processes. Lead is known to interfere with synaptic function by disrupting neurotransmitter release and receptor signaling. Specifically, Pb exposure can alter the activity of voltage-gated sodium channels and inhibit acetylcholinesterase (AChE), leading to the accumulation of acetylcholine at synapses. This disruption in cholinergic transmission, along with interference in GABAergic signaling pathways, compromises synaptic plasticity and neuronal communication—critical processes for learning and memory [4], [22]. Notably, a review by Maleki et al. reported that maternal exposure to environmental pesticides is associated with increased ASD progression risk in children. Their review identified key mechanisms—including reactive oxygen species (ROS) generation, prostaglandin E2 synthesis, AChE inhibition, voltage-gated sodium channel disruption, and GABA inhibition—as potential contributors to neurodevelopmental disturbances leading to ASD [25]. Although their primary focus was on pesticides, these mechanistic pathways share striking similarities with those observed in heavy metal toxicity. For example, both lead and pesticides can induce oxidative stress and disrupt cholinergic and GABAergic signaling, suggesting a possible common pathway by which environmental exposures might contribute to ASD risk. Moreover, genetic and epigenetic factors may modulate an individual’s susceptibility to Pb toxicity, altering both the extent of heavy metal accumulation and its neurodevelopmental impact.

Thus, our findings on Pb in hair samples, while paradoxical at first glance, could be explained by variations in genetic determinants of metal metabolism or by adaptive physiological responses in populations chronically exposed to environmental pollutants. These considerations underscore the importance of further research that integrates environmental, genetic, and epigenetic data to fully elucidate the complex interplay of factors influencing ASD pathogenesis.

Cadmium’s null result, while not ruling out potential long-term risks, suggests that Cd may not be a primary metal of concern for ASD etiology in this particular region. Local sources of cadmium, such as certain industrial emissions, may be less prominent in Makassar, or families in both groups may similarly avoid significant Cd exposure [11], [32]. In contrast, family history of ASD standing out as a potent risk factor underlines the established roles of genetics and shared environmental settings [16], [24]. When coupled with prenatal tobacco smoke exposure, it underscores the complexity of neurodevelopmental outcomes, in which modifiable and intrinsic factors intertwine [6], [13], [21].

3.4. Limitations and directions for future research

Despite its important contributions, this study has several limitations that should be considered when interpreting the results. First, the cross-sectional design restricts our ability to infer causality between heavy metal exposures and ASD; longitudinal studies are needed to establish temporality and causal relationships. Second, while the use of hair samples provides a noninvasive method to estimate chronic exposure, hair metal concentrations can be influenced by external contamination, individual hair characteristics, and local environmental factors, which may introduce measurement variability. Third, the sample size is relatively modest (n = 60), potentially limiting the statistical power to detect small effects and increasing the risk of type II errors. Fourth, the study design involved an unmatched case-control approach. Although we statistically adjusted for multiple confounders (such as age, gender, family income, history of exclusive breastfeeding, family history of ASD, maternal dietary habits, cosmetic use, and tobacco smoke exposure), residual confounding cannot be entirely ruled out due to potential inaccuracies in self-reported data and other unmeasured variables. Finally, the reliance on parental questionnaires introduces the possibility of recall bias, particularly regarding exposures such as exclusive breastfeeding duration, maternal consumption patterns, and tobacco exposure.

Despite these limitations, this study offers several strengths that support the validity and relevance of its findings. One key strength is its focus on an urban Indonesian population—a setting that is underrepresented in the literature but is characterized by unique environmental exposures resulting from rapid industrialization and local cultural practices. The integration of advanced analytical techniques, such as inductively coupled plasma–mass spectrometry (ICP–MS), has enabled precise quantification of hair heavy metal concentrations, thereby providing robust exposure assessment. Moreover, the study’s comprehensive evaluation of potential confounders through extensive bivariate analyses, coupled with the application of the Boruta algorithm for variable selection, enhances the reliability of our multivariable logistic regression models. This rigorous statistical approach minimizes the impact of confounding and improves the ability to isolate the independent associations between heavy metal exposures and ASD risk.

Additionally, by evaluating both direct (heavy metal concentrations) and indirect (maternal dietary and behavioral patterns) exposure routes, the study provides a more holistic understanding of the factors influencing neurodevelopment. The inclusion of variables such as maternal use of lightening creams and prenatal tobacco exposure enriches the analysis by linking environmental toxicant exposure to potential risk-modifying behaviors. Finally, the study's approach of integrating environmental exposure assessment with established genetic and epigenetic considerations in the discussion offers valuable insights into possible mechanisms underlying heavy metal toxicity, particularly regarding lead’s impact on synaptic function and neurotransmission. These strengths collectively enhance the study’s contribution to the field and underscore the potential for targeted public health interventions in similar urban contexts.

4. Conclusion

This study highlights the complex interplay of genetic and environmental factors in ASD among school-aged children in Makassar. Elevated hair mercury emerged as a principal risk factor, reflecting maternal behaviours—such as frequent seafood consumption and the use of mercury-containing cosmetic products. In contrast, lead levels were unexpectedly lower in the ASD group, indicating a paradoxical finding that may reflect unique genetic or epigenetic variations affecting lead metabolism and clearance in this population. Cadmium did not significantly differ between children with and without ASD.

Additionally, important maternal factors—specifically the use of skin-whitening creams and prenatal tobacco smoke exposure—along with a positive family history of ASD, were significantly associated with higher ASD risk. These findings provide compelling scientific evidence that both environmental exposures and familial predispositions play crucial roles in the multifactorial etiology of ASD. The observed associations emphasize the necessity for targeted public health strategies aimed at reducing exposure to neurotoxicants and mitigating risk factors during critical periods of neurodevelopment in vulnerable populations.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Institute for Research and Community Services, Hasanuddin University [grant number 00323/UN4.22/PT.01.03/2023].

CRediT authorship contribution statement

Aris Ahmad Zaharin: Writing – review & editing, Formal analysis. HK Adhariana: Investigation, Data curation. Yusbud Mahfuddin: Investigation, Data curation. Natsir Muhammad Fajaruddin: Investigation, Data curation. Susilawaty Andi: Writing – review & editing, Investigation. Amqam Hasnawati: Writing – original draft, Supervision, Project administration, Investigation, Formal analysis, Conceptualization. La Ane Ruslan: Writing – review & editing, Writing – original draft, Methodology, Formal analysis, Data curation, Conceptualization.

Author contribution

RLA, HA, AHK and AS have been involved in research concepts and designs. NMF, YM, HDN and DVA have been involved in data collection. RLA, HA and AS analyzed the data and wrote the manuscript. AZA reviewed and approved the final manuscript.

Declaration of Generative AI and AI-assisted technologies in the writing process

During the preparation of this work the authors used ChatGPT 4.0 in order to extract metadata from collected articles used in citation and references. After using this tool, the authors reviewed and edited the content as needed and takes full responsibility for the content of the publication.

Declaration of Competing Interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Ruslan La Ane reports financial support was provided by Hasanuddin University. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

Data will be made available on request.