Association of metallic elements with telomere length in children with autism spectrum disorder

Study design and population

Eighty-three children diagnosed with ASD at the Maternal and Child Health Hospital of Longgang District, Shenzhen, between November 2021 and December 2023, and 95 typically developing children from the same period, were selected as case-control subjects for this study. The final diagnosis of ASD was made according to the criteria in the Diagnostic and Statistical Manual of Mental Disorders-5 (DSM-5), Fifth Edition, by two specialists with over three years of clinical experience, in conjunction with the developmental history of the children. Children with chromosomal or inherited metabolic disorders, other neurodevelopmental disorders, a history of psychotropic medication use, or those who had recently taken micronutrient supplements were excluded from the study population. The study was approved by the ethics committee of Maternal and Child Health Hospital of Longgang District, Shenzhen, China under the approval number LGFYYXLLL-2021-003. Written informed consent was obtained from the primary guardians of all participating children.

Assessment of the child’s basic condition and autism symptoms

The questionnaire collected data on socio-demographic information, basic details about the child, and parents’ behavioral habits, which was completed by the child’s caregiver. All children were screened using the Child Autism Rating Scale (CARS) and the Autism Behavior Checklist (ABC). The CARS scale (Xiao et al., 2022) assesses children with ASD in 15 items, including interpersonal relationships, imitation (of words and actions), emotional responses, and somatic application skills. Each item is rated on a scale from 1 to 4, with a total score of 30 or greater indicating a diagnosis of autism, less than 36 indicating mild to moderate autism, and a score of 36 or greater indicating severe autism. The ABC scale (Wei et al., 2021) consists of 57 items grouped into five sections: sensory, relating, body and object use, language, and social and self-help skills. Each item is scored on a scale from 1 to 4, with a total score of <53 considered negative for screening, a score between 53 and 67 considered positive for screening, and a score of ≥68 used to aid in the diagnosis of autism. A higher total score indicates more severe behavioral symptoms of autism.

Detection of metal elements and TL

Inductively coupled plasma mass spectrometry (ICP-MS) was used to detect six metallic elements in children’s whole blood: Mn, Cu, Zn, Ca, Mg and Fe. The whole blood specimen was placed on a blood mixer for thorough mixing, and 0.1 mL of blood was accurately measured into 1.9 mL of diluent (0.1% HNO₃ + 0.1% X-100), and mixed for 30 s before testing.

Genomic DNA extraction from peripheral blood mononuclear cells (PBMCs). The lymphocyte layer was isolated from 1 mL of peripheral blood using a lymphocyte isolation solution within 2 h after venous blood collection from children, conducted by a pediatric nurse using a 5-mL EDTA anticoagulation tube (Solebo, Beijing, China). The white membrane layer between the plasma and isolate was aspirated with a pasteurized pipette, and the lymphocytes were washed with PBS and resuspended to obtain the final lymphocyte suspension. Cellular DNA was extracted using magnetic bead separation on a Lab-Aid 820 Nucleic Acid Extractor (Xiamen Zhishen Biotechnology Co., Ltd., Xiamen, China). The quality and concentration of the extracted DNA samples were evaluated using a Qubit fluorometer, and the DNA was stored at −80 °C until further analysis. Employing Digital Polymerase Chain Reaction (dPCR) (Beijing Xinyi Biological Co., Ltd., Beijing, China) absolute quantitative TL. dPCR is a third-generation PCR technique developed from traditional PCR and real-time fluorescence quantitative PCR (qPCR). This method enables the direct quantification of TL in target sequences without requiring standard dilution curves, reaction efficiency calibrations, or reference samples, offering high accuracy and precision. Its distinct advantages in the absolute quantification of DNA molecules have led to widespread adoption in various precision medicine applications, with its validity supported by recent literature (Quan, Sauzade & Brouzes, 2018). To further ensure the accuracy of our experimental results, we incorporated an internal reference gene as a control. Primer designs were adapted from prior studies (Gao et al., 2020). The target gene was TL, and the internal reference gene was Hemoglobin Subunit Beta (HBB); the primer sequences were as follows: TL-Forward: 5′-CGGTTTGTTTGGGTTTGGGTTTGGGTTTGGGTTTGGGTT-3′, TL-Reverse: 5′-GGCTTGCCT TACCCTTACCCTTACCCTTACCCTTACCCT-3′, HBB-Forward: 5′-AGGAGAAGTCTGCCGTTACTG-3′, HBB-Reverse: 5′-CCGAGCACTTTCTTGCCATGA-3′. The reaction system was 30 µL, with the reaction conditions as follows: 95 °C for 3 min, 95 °C for 15 s, 55 °C for 20 s, 72 °C for 25 s, and 40 cycles.

Statistical analyses

Epidata 3.1 software was used for data entry and database creation, and R version 4.3.3 (R Core Team, 2024) was applied for statistical analysis. Categorical variables were summarized by the number of cases and composition ratio, while continuous variables that conformed to a normal distribution were represented as the mean ± standard deviation $(\overline{\mathrm{x}}\pm \mathrm{s})$. Continuous variables not following a normal distribution were reported as M (P₂₅, P₇₅). Chi-square tests were conducted to compare the group differences on categorical variables, while independent samples t-test and the Mann-Whitney U test were conducted to compare continuous variables that underwent normality testing.

Linear regression analysis was employed to assess the effect of metal content on TL in both groups of children, using standardized TL as the dependent variable and the content of the six metallic elements as the independent variables. Based on previous literature and the results of univariate analyses, the following relevant covariates were included in our model, including children’s age, height, weight, sex, outdoor exercise time, family income, balanced dietary status, and parental smoking status. Additionally, we investigated the dose-response relationships and combined exposure effects of six metal elements and TL levels in children with ASD using the Bayesian kernel machine regression (BKMR) model, implemented with the BKMR package in R. TL served as the outcome variable, the six metal elements were treated as a mixture of exposures, and the covariates described above were used as well. The model was run for 10,000 iterations. The dose-response relationship for combined exposures was plotted, and posterior selection probabilities and effect values for each exposure were estimated. The dose-response curves were plotted for each exposure, and a final curve was plotted for first-order interactions within the mixed exposure. The variable q.fixed represents a specific percentile of other mixed exposures fixed during exposure calculations, while est provides the overall risk estimate of the mixed exposures. Posterior inclusion probability (PIP) indicates the relative importance of each exposure, with values closer to one representing higher importance. Significance was set at P < 0.05.

Sample characteristics

No statistically significant difference was observed in the mean age at entry between the two groups (t = 0.67, P = 0.502). The mean TL were 5,109.11 ± 2,028.10 kb and 6,230.1 ± 2,931.73 kb for each group, respectively, with a statistically significant difference between groups (t = 3.00, P = 0.003). No statistically significant differences were observed in height, weight, gender, ethnicity, birth weight, parental occupation, or daily outdoor playtime between groups (P > 0.05), However, statistically significant differences were identified in levels of Mn, Cu, Ca and Mg (P < 0.05), while no significant differences were found for Zn or Fe (P > 0.05) (Table 1). Compared to the typically developing group, the ASD group exhibited shorter TL, higher concentrations of Mn and Mg, and lower concentrations of Cu and Ca.

The CARS scale and ABC scale scores are shown in Table 2. The mean CARS scale score of 32.95 ± 4.76 suggests that children with ASD exhibit mild to moderate symptom severity. The mean ABC scale score of 54.59 ± 15.07 meets the positive diagnostic criterion, confirming the presence of ASD-related symptoms. The sensory score of 7.81 ± 4.54 indicates that children with ASD experience some mild sensory processing difficulties. The high relating score of 16.09 ± 5.76 reflects significant social interaction challenges, a core feature of ASD. Body and object use scores (6.91 ± 6.43) were relatively low but showed variability, suggesting that while some children with ASD may experience mild motor skill difficulties, not all have significant issues. The language score of 14.09 ± 5.68 was high, highlighting notable language development challenges, a common characteristic of autism. Social and self-help skills scores of 9.95 ± 4.62 were moderate, suggesting variability in self-care abilities, with some children encountering general challenges. Overall, children with ASD exhibited mild to moderate symptom severity, with pronounced difficulties in social interaction and language development. Additionally, difficulty levels varied across sensory processing, body and object use, social and self-help skills.

Correlation between metallic elements and TL in two groups of children

Ca showed a protective effect against TL in the ASD group (β = 0.07, 95% CI [0.01–0.13], P = 0.027), while Mg had a similar effect in the control group (β = 0.10, 95% CI [0.01–0.18], P = 0.027). The variance inflation factor (VIF) indicated no covariance among the six metallic elements in the model (Table 3).

Combined effect of metal exposure on TL in the BKMR model in ASD group

The BKMR model was employed to quantify the combined influence of six metallic elements on TL in individuals with ASD. The conditional posterior inclusion probabilities (condPIP) for each metal, as a measure of their involvement, are shown in Table 4. The findings suggest that Ca may play a pivotal role in the composite effects of multiple metal exposures on TL. Notably, the model revealed a statistically significant positive effect of the metal mixture on TL. Compared to the median (50th percentile) exposure level, significant combined effects were observed at alternative percentiles, as depicted in Fig. 1A.

When concentrations of other metals were held constant at their 25th, 50th, and 75th percentile values, a pronounced negative correlation emerged between changes in Zn and Mn concentrations (spanning the 25th to 75th percentiles) and TL (Fig. 1B). Conversely, a significant positive correlation was noted between changes in Cu, Ca, Mg, and Fe concentrations across the same percentile range and TL (Fig. 1B). To assess potential nonlinearity in the exposure-response relationship for each metal in isolation, univariate analyses were conducted while holding other exposures at their median levels. The results revealed positive exposure-response associations between TL and the levels of Cu, Ca, Mg, and Fe (Fig. 1C). In contrast, inverse exposure-response relationships were found between TL and Mn and Zn levels (Fig. 1C). No significant interactions among metals in their effects on TL were observed (Fig. 1D).