Circulating thyroid hormones and metabolites in children with autism spectrum disorder

Context: Thyroid hormones affect neurological development and function, but detailed studies of thyroid hormones and metabolites in autism are lacking.

Objective: To characterize thyroid function and metabolism in autistic children.

Methodology: This cross-sectional study compared 788 autistic children (mean age 7.6 ± 3.9 years, 78% male) with 301 non-autistic children (mean age 7.8 ± 4.0 years, 48% male; comprising 215 (71.4%) non-autistic siblings of participants and 86 (28.6%) unrelated individuals). Plasma TSH, free T4 (FT4) and free T3 (FT3) were measured by automated immunoassay, and total T4, total T3 and thyroid hormone metabolites by customized liquid chromatography-tandem mass spectrometry (LCMS/MS). Regression analyses were adjusted for age and sex.

Results: TSH concentrations were similar in autistic and non-autistic children (median 2.3 vs 2.1 mU/L, P = 0.64). FT4 was significantly lower in autistic children (18.4 vs 18.7 pmol/L, P = 0.0003), as was FT3 (7.0 vs 7.1pmol/L, P<0.0001), with no significant difference in the FT4:FT3 ratio (P = 0.24). Total T4 was lower in autistic children (178 vs 194 nmol/L, P = 0.0026, as was total T3 (2.2 vs 2.4 nmol/L, P = 0.018), with no significant difference in the T4:T3 ratio (P = 0.099). Two metabolites were significantly lower in autistic children: 3,5-T2 (0.010 vs 0.021 nmol/L, P<0.0001) and 3,3’-T2 (0.12 vs 0.16 nmol/L, P<0.0001), whereas T0 levels were higher (1.5 vs 1.1 nmol/L, P = 0.028).

Conclusions: Circulating thyroid hormones and metabolites differ between autistic and non-autistic children, although the observed differences are small. The study demonstrates the utility of LCMS/MS for in-depth characterization of thyroid hormone economy, with potentially wide applications.

Introduction

Autism spectrum disorder (autism, ASD) refers to a neurodevelopmental disability characterized by qualitative and lifelong differences in social interaction and communication as well as the presence of repetitive and sensory behaviors and interests (1). Most countries with epidemiological data have reported a marked increase in autism diagnoses over recent decades, with current global prevalence estimates of 1.2% to 2%, and a male to female ratio of approximately 3:1 (2, 3). Although the biological pathways contributing to autism remain unclear, the current consensus is for a multifactorial etiology, incorporating a constellation of genetic risk variants interacting with environmental factors (4, 5).

Circulating thyroid hormones have essential roles in brain development and function and are important for childhood development and adult health (6). They have a critical role in in the development of the central nervous system during embryogenesis and early infancy, essential in neurogenesis and oligodendrogenesis, cerebral cortex maturation and myelination (6, 7). The crucial roles of thyroid hormone are demonstrated by the adverse effects of neonatal hypothyroidism on multiple regions and pathways, the most sensitive being cerebral cortex, hippocampus, cerebellum and corticospinal tracts (6).

Under the influence of thyroid stimulating hormone (TSH), the major hormone secreted by the thyroid is thyroxine (T4) which serves largely as a prohormone, being converted to the biologically active hormone triiodothyronine (T3) in peripheral and target tissues. Thyroid hormone biosynthesis, metabolism and regulation are highly complex, far more so than the well described relationships between free T3 (FT3), free T4 (FT4) and TSH. The best established metabolic pathways for thyroid hormones are catalyzed by the iodothyronine deiodinase enzymes D1, D2 and D3, which in a tissue-specific manner convert T4 to T3 and reverse T3, with further deiodination to diiodothyronines, monoiodothyronines and L-thyronine (8, 9). In addition, there are alternate metabolic pathways of decarboxylation to iodothyronamines and oxidative deamination to iodothyroacetic acids. Notably, some metabolites are thought to be biologically active, having been shown to activate nuclear thyroid hormone receptors (8). Tissue concentrations of T3 within the brain are tightly controlled, derived in part from circulating T3, and also from D2-mediated deiodination of T4 (7). In glial cells and certain interneurons, D2 converts T4 to T3, whereas D3 converts T4 and T3 to the inactive metabolites rT3 and T2 respectively (7). The balance between D2 and D3 activity during development is essential to ensure critical neural tissues are exposed to adequate concentrations of T3.

Varying degrees of thyroid dysfunction have been associated with neurodevelopmental and psychiatric phenotypes. Overt and clinically diagnosed thyroid dysfunction are associated with neuropsychological dysfunction, including mood, behavior, emotion, and cognition (10). Variation in thyroid function, even within the normal range, is also associated with impaired mood (10) and cognitive decline in the elderly (11–13), as well as diagnoses of depression (14) and attention deficit hyperactivity disorder (ADHD) (15, 16). In a recent meta-analysis, circulating free T4 concentrations were reported to be reduced in children with neurological disorders (including ADHD, autism and tic disorder) compared with controls, whereas TSH levels were not significantly different (17). In the developing fetus, transplacental passage of maternal thyroxine plays an important role in fetal brain development during the first and second trimester, until the fetal hypothalamo-pituitary-thyroid axis is functionally mature (18). In longitudinal studies, infants exposed in utero to either maternal hypothyroidism or thyrotoxicosis are more likely to experience developmental difficulties in motor, social, emotional and language domains (19). Further studies have identified an inverted U-shaped association of higher and lower levels of maternal FT4 during pregnancy (excluding mothers with overt thyroid disease) and reduced cognitive abilities in offspring, as well as lower grey matter and cortex volume (20).

The physiological role of thyroid hormones in brain development makes an association between variation in thyroid function and the biological pathways contributing to autism plausible. Tentative links between thyroid hormone exposure and autism were made in 1970, with suggestions autistic phenotypes may be as a result of delayed diagnosis of thyroid dysfunction (21). Considering this, a small randomized controlled trial with crossover design of 30 clinically euthyroid autistic children was performed, with supraphysiological doses of T3 having no overall benefit on a range of behaviors associated with autism, although there was apparent benefit in a subgroup with lower cognitive abilities (22). These results have not been replicated. Subsequent observational studies were of low quality, with an autistic population different to those in recent clinical practice, and with significant methodological limitations including small sample size and post-mortem tissue analysis (21, 23–25). More recent studies report an increased risk of autism diagnosis in children with a family history of autoimmune thyroid disease (26), and maternal thyroid dysfunction during pregnancy, particularly maternal hypothyroxinemia, has been associated with increased likelihood of autism diagnosis in offspring (15, 27–33). Animal studies have provided evidence that maternal hypothyroxinemia can induce an autism-like phenotype in offspring (34). However, the findings in this area remain inconclusive, and may be affected by misclassification, survival and publication biases (28).

Despite this previous work, there are few detailed studies of thyroid function in autism. A recent case-control study examining thyroid hormones in children with neurological disorders reported altered thyroid parameters, with lower FT4, higher FT3, and lower TSH in children with autism compared with controls (17). To date, there are no data regarding thyroid hormone metabolites in children with autism. We conducted a study of participants in the Australian Autism Biobank, hypothesizing that differences in thyroid hormones and/or metabolites may be present in autistic children compared with non-autistic individuals, and within the ASD group, might be associated with clinical phenotype.

Methods

A cross-sectional, observational study was performed, examining thyroid hormones and metabolites in autistic children (ASD group) and non-autistic comparators (non-ASD). Data came from the Australian Autism Biobank (AAB), which collected phenotypic data and biological samples from four autism research centers or developmental clinics across Australia (Perth, Brisbane, Melbourne, Sydney), as previously described (35). The ASD group consisted of children with a clinically confirmed diagnosis of autism per the Diagnostic and Statistical Manual of Mental Disorders, 4^th and 5^th edition (DSM-IV, DSM-5) criteria. The non-ASD group included children with no known developmental differences, comprising non-autistic siblings of autistic participants and unrelated children with no family history of autism.

Participants were excluded if they had a history of thyroid disease, were taking thyroid hormones, antithyroid drugs or medications known to affect thyroid function tests (e.g. lithium, amiodarone, carbamazepine, phenytoin, quetiapine, oral contraceptive and systemic glucocorticoids), had biochemical evidence of clinically significant thyroid dysfunction or results which were outliers (TSH <0.1 or >10 mU/L; FT4, FT3, TSH outside mean ± 4 SD limits).

Phenotypic data

The AAB contains detailed phenotypic data, including sociodemographic data, family history information, and several clinical assessments completed by trained researchers to gauge behavioral presentation and cognitive abilities.

Autistic behavioral characteristics

The Autism Diagnostic Observation Schedule, 2^nd Edition (ADOS-2) (36) was administered to the participants in the autism group only. The ADOS-2 is a semi-structured assessment that uses simple activities and questions to elicit and observe communicative and social behaviors relevant to the diagnosis of autism. A range of ADOS-2 modules are available that allows tailoring of activities to the level of verbal language. The ADOS-2 calibrated severity score (range, 1–10 points) was developed to facilitate comparison across different developmentally staged ADOS-2 modules (37). Higher ADOS-2 calibrated severity scores represent greater manifestation of autism behavioral characteristics.

Cognitive abilities

One of two measures of cognitive ability was administered to participants, depending on age. Children aged between 2 and 6 years of age were administered the Mullen Scales of Early Learning (MSEL), a standardized developmental assessment involving interactive and play-based tasks (38). Four domains were assessed: fine motor, visual reception, expressive language and receptive language. Children aged above 6 years of age were administered the Wechsler Intelligence Scale for Children 4th edition (WISC-IV) (39), which involves ten structured activities eliciting cognitive abilities in four domains: verbal comprehension, perceptual reasoning, working memory and processing speed. Intellectual disability (ID) was defined as MSEL < 70 or WISC-IV < 70.

Functional abilities

The Vineland Adaptive Behavior Scale, 2^nd edition (VABS–2) is a parent-report questionnaire that provides a measure of child adaptive behavior (40). The VABS-2 assesses four domains: communication, daily living skills, socialization and motor skills (for children under 6 years of age), which sum together to give an Adaptive Behavior Composite (ABC), standardized around a mean of 100 and standard deviation of 15. A higher ABC indicates greater functional ability. Functional impairment (FI) was defined as VABS-2 ABC < 70.

Biological sample collection

Venous blood samples were collected from participants under non-fasting conditions by trained pediatric phlebotomists or through hospital/pathology phlebotomy services. Time of blood collection was not standardized, and occurred during office hours (range 0705–1700 h); the median time of collection was 1230 h (inter-quartile range 1149 h, 1330 h). Samples were transported to the University of Queensland’s Institute for Molecular Bioscience for processing, then stored at -80C. Samples were then sent on dry ice to Queen Elizabeth II Medical Centre in Perth for hormone and metabolite analysis. Routine measures of thyroid function were generated from both serum and plasma samples by automated immunoassay using the Roche Cobas e platform (Roche Diagnostics, Australia), measuring FT3, FT4, TSH and thyroid peroxidase antibodies (TPOAb).

Custom developed liquid chromatography-tandem mass spectrometry (LCMS/MS) assays were set up to measure total T4, total T3 and 11 thyroid hormone metabolites (3,3’,5’-triiodothyronine (rT3), 3,3’-diiodothyronine (3,3’-T2), 3,5-diiodothyronine (3,5-T2), 3,5-diiodothyronamine (T2AM), 3 - iodothyronamine (T1AM), 3-iodothyronine (T1), L-thyronine (T0), tetraiodothyroacetic acid (T4A/tetrac), triiodothyroacetic acid (T3A/triac), diiodothyroacetic acid (T2A) and 3-iodothyroacetic acid (T1A)), according to guidelines set out by the Clinical and Laboratory Standards Institute (CLSI C57-ED1) (41). Validation included assessment of limit of detection (LOD), limit of quantification (LOQ), linearity, recovery, matrix effects, specificity, intra-assay repeatability, total analytical repeatability, reference values, robustness, carry-over and sample stability. Full details of assay methodology, validation and performance are provided as Supplementary material, including Supplementary Tables 1-5. Calibration standards were made from certified reference material obtained from various suppliers as listed in Supplementary Table 3. Acetonitrile protein crush method was used for sample extraction. Fifty microliters of calibrator, quality control and patient samples were spiked with 25 µl of internal standards mixed with 200 µL of 90% acetonitrile 10% methanol solvent solution. After vortexing and centrifugation, 220 µL of supernatant from each sample was collected and dried down in a vacuum concentrator. Dried samples were reconstituted in a solution of 50% of methanol and 50% water. Ten microliters of sample was injected for LCMS for analysis.

LC-MS/MS analysis was performed by a Waters Xevo TQ-XS tandem mass spectrometer coupled with a Waters ACQUITY^® I-Class UPLC system with binary pumping capability (Waters Corporation, Milford, Massachusetts, USA) (Supplementary Table 1). Experimental setup was programmed by Waters software MassLynx and data was processed with Waters software TargetLynx. Analyte ion detection was performed with positive and negative electrospray ionization in multiple reactions monitoring (MRM) mode (Supplementary Table 2). All analytes had acceptable low instrument level of detection for the purpose of the method, acceptable linearity (defined as r2>0.995) over relevant concentration ranges and acceptable level of matrix effects (defined as 100 ± 30%). Assay imprecision was measured by the performance of 3 in-house quality control samples for each analyte; intra-day precision ranged from 2.1-8.0% and inter-day precision from 3.2-14.9% (Supplementary Table 5). At the time of establishing the assays (early 2021), no external quality assurance program was available for total T4 and T3 by LCMS/MS; quality control was assured by use of certified reference materials directly to make up standards in methanol with low dilution factors, together with added internal standard in each sample to justify for any sample recovery variance, ion suppression and instrument sensitivity fluctuation.

All 13 analytes were successfully detected on spiked sera (Supplementary Figure 1), whereas in participant sera total T4, total T3, L-thyronine (T0), 3,3’,5’-triiodothyronine (reverse T3; rT3), 3,3’-diiodothyronine (3,3’-T2) and 3,5-diiodothyronine (3,5-T2) were detected. The remaining thyroid hormone metabolites were below the limit of detection in all participant samples: 3-iodothyronine (T1), 3-iodothyronamine (T1AM), 3,5-diiodothyronamine (T2AM), 3-iodothyroacetic acid (T1A), diiodothyroacetic acid (T2A), triiodothyroacetic acid (T3A/triac) and tetraiodothyroacetic acid (T4A/tetrac).

Statistical analysis

Descriptive statistics for continuous variables are given as median and quartiles, and between group comparisons performed using the Mann Whitney U test. Categorical variables are presented as proportions and compared using the Chi-squared or Fisher’s exact test. Skewed variables (all except FT3 and FT4) were then square root or log transformed, as required, before modelling of the mean via regression analysis. Thyroid hormones and metabolites were compared between the ASD and non-ASD groups via linear mixed effects models, which accounted for family relatedness, and adjusted for the covariates age and sex. A generalised linear mixed effects model was used for TPOAb, which was treated as a binary variable using the manufacturer’s recommended reference range of <35 IU/L. The following derived indices of thyroid hormone metabolism and pituitary-thyroid axis function were calculated and compared between ASD and non-ASD: FT4/FT3 ratio, T3/rT3 ratio, FT4xTSH product and TSH index (ln TSH + 0.1345 x FT4) (42, 43).

In the group of autistic children, we assessed associations between thyroid hormones and metabolites and the level of autistic behaviors (using the calibrated severity score of the ADOS-2). Comparisons of analyte levels were made between non-autistic children and (a) autistic children with (ASD+ID) and without (ASDnoID) intellectual disability and (b) autistic children with (ASD+FI) and without (ASDnoFI) functional impairment, as defined above.

Two sensitivity analyses were performed. Firstly, thyroid hormone and metabolite levels were compared between autistic children and each of non-autistic siblings and unrelated controls, as well as between siblings and unrelated controls. Secondly, to examine the potential impact of neurosyndromic diagnoses, such as major chromosomal anomalies and neurological disorders established to be relevant to autism, analyses were performed after excluding children with such diagnoses.

The significance threshold was set at 5% throughout, with false discovery rate adjustment using the Benjamini-Hochberg procedure when multiple comparisons were made in the same analysis. Statistical analysis was performed in the R statistical computing environment, version 4.3.2 (44).

Ethics approval

The study was conducted in accordance with the Declaration of Helsinki, and approved by the Human Research Ethics Committee of the University of Western Australia (protocol code 2020/ET000102, approved 9th March 2021). Informed consent was obtained from all participants and/or their legal guardians.

Results

A total of 788 autistic (mean age 7.6 ± 3.9 years, 78% male) and 301 non-autistic children (mean age 7.8 ± 4.0 years, 48% male) were included for analysis, after the exclusion of 19 participants due to a history of thyroid disease, use of thyroid hormones or thyroid-altering medications or biochemical evidence of clinically significant thyroid dysfunction (TSH < 0.1 and > 10 mU/L) (Figure 1). Of the 301 non-ASD children, 215 (71.4%) were non-autistic siblings of participants and 86 (28.6%) were unrelated. As expected, the ASD group contained a greater proportion of males and had a higher prevalence of intellectual and functional impairment than the non-ASD group, and autistic children were more likely to be taking prescribed medications (Table 1).

Figure 1

Figure 1. Participant flowchart.

Table 1

Table 1. Characteristics of the ASD and non-ASD groups.

Thyroid function testing by immunoassay revealed statistically significant differences between autistic and non-autistic children, with lower FT4 (median 18.4 vs 18.7 pmol/L, P = 0.0003) and FT3 (7.0 vs 7.1 pmol/L, P<0.0001) after adjustment for age and sex (Table 2). In each case, the magnitude of the difference between groups was small, with median values well within the reference range. TSH, TPOAb and the FT4:FT3 ratio did not differ significantly between groups (P = 0.64, 0.31 and 0.24 respectively).

Table 2

Table 2. TSH, FT4, FT3, FT4:FT3 ratio and TPOAb in ASD and non-ASD groups, measured by immunoassay.

Total T4 as measured by LCMS/MS was lower in autistic children compared with non-autistic (178 vs 194 nmol/L, P = 0.0026), as was total T3 (2.2 vs. 2.4 nmol/L, P = 0.018), with no significant difference in the T4:T3 ratio (P = 0.099) (Table 3). Reverse T3 concentrations did not differ between groups, whereas the T3:rT3 ratio was lower in autistic children compared with non-autistic (2.9 vs 3.1, P = 0.003). Two metabolites had lower concentrations in autistic children compared with non-autistic: 3,5-T2 (0.01 vs 0.021 nmol/L, P<0.0001) and 3,3’-T2 (0.12 vs 0.16 nmol/L, P<0.0001), whereas levels of T0 were higher in ASD than non-ASD (1.5 vs 1.1 nmol/L, P = 0.028).

Table 3

Table 3. Thyroid hormone and metabolite levels in ASD and non-ASD groups, measured by LCMS/MS.

Of the remaining derived indices, the TSH index did not differ significantly between groups (P = 0.07), nor did the FT4xTSH product (P = 0.63).

Within the group of autistic children, there were no significant relationships between the ADOS-2 calibrated severity score, and any of the thyroid hormones and metabolites (Table 4).

Table 4

Table 4. Relationships between thyroid hormone and metabolite levels with ADOS-2 calibrated severity score in autistic children.

In subgroups of ASD with and without intellectual disability, results appeared similar to those for the group as a whole (Figure 2). FT4, FT3, 3,3’-T2 and 3,5-T2 levels were significantly lower in both subgroups than the non-ASD group. In the subgroup with intellectual disability (37% of cases), total T4 and total T3 were no longer significantly different from non-ASD, and T0 did not differ significantly from non-ASD in either subgroup, probably reflecting reduced statistical power.

Figure 2

Figure 2. Thyroid hormones and metabolites, measured by immunoassay and LCMS/MS, in autism cases with and without intellectual impairment (ASD+ID and ASDnoID respectively) compared with non-ASD. Data are median, upper/lower quartile and highest/lowest value. *P < 0.05, **P < 0.01, ***P < 0.001.

In subgroups of ASD with functional impairment and those without, results appeared broadly similar to those for the cohort as a whole, but differences in FT4 and total T3 were no longer significant compared with non-ASD (Figure 3). Levels of 3,3’-T2 and 3,5-T2 levels were significantly lower, and T0 higher in both subgroups than non-ASD (consistent with the entire group), whereas in children with functional impairment (41% of cases), FT3 and total T4 levels did not differ significantly from non-ASD.

Figure 3

Figure 3. Thyroid hormones and metabolites, measured by immunoassay and LCMS/MS, in autism cases with and without functional impairment (ASD+FI and ASDnoFI respectively) compared with non-ASD. Data are median, upper/lower quartile and highest/lowest value. *P < 0.05, **P < 0.01, ***P < 0.001.

In a sensitivity analysis comparing thyroid hormone and metabolite levels between autistic children, non-autistic siblings and unrelated controls, results were similar to those for the primary analysis. Demographic data are shown in Supplementary Table 6 and results in Table 5. For analytes which differed significantly between ASD and non-ASD in the primary analysis, the direction of the effect was the same in the sensitivity analysis, although the magnitude and significance varied with the smaller sample size, and some comparisons were no longer statistically significant. The only analyte which differed significantly between siblings and unrelated controls was T0, being lower in siblings (1.0 vs. 1.3 nmol/L, P = 0.0006), and not significantly different between ASD cases and unrelated controls. In a second sensitivity analysis, results were essentially unchanged after exclusion of 16 autistic children with established neurosyndromic diagnoses relevant to autism, including major chromosomal anomalies and neurological disorders.

Table 5

Table 5. Sensitivity analysis comparing thyroid hormones and metabolites in ASD (N = 788) versus each of non-autistic siblings (N = 215) and unrelated controls (N = 86), and between siblings and unrelated controls.

Discussion

The pathogenesis of autism is poorly understood, but is thought to involve interactions between genetic and environmental factors (4, 5). Thyroid hormones have critical roles in brain development and function (45–47), making it biologically plausible that altered thyroid hormone economy could contribute to the pathogenesis of autism. Thyroid hormone metabolism is complex and tissue-dependent, and although T3 is the most biologically active hormone, metabolites such as 3,5-T2 are also of interest, as they can bind to and activate nuclear thyroid hormone receptors (8), and may have other effects including altered mitochondrial function through non-genomic mechanisms (8, 48, 49).

The aim of this study was to test the hypothesis that alterations in circulating thyroid hormones or metabolites are present in autistic children compared with non-ASD, and might be relevant to autism pathogenesis. Although TSH did not differ significantly between groups, circulating concentrations of total T4 and T3 as well as free T4 and T3 were lower in the ASD than the non-ASD group, accompanied by reduced levels of the thyroid hormone metabolites 3,5-T2 and 3,3’-T2 and higher levels of T0, whereas reverse T3 did not differ significantly between groups. There were no significant relationships between thyroid parameters and autism severity (as assessed by ADOS-2 calibrated severity score). Results appeared consistent across subgroups of ASD with and without intellectual disability or functional impairment, although for some analytes, differences between subgroups and non-ASD were no longer significant, probably reflecting reduced statistical power. Although statistically significant, the magnitude of the differences between ASD and non-ASD groups was small, and likely to be within the bounds of biological variation and uncertainty of measurement. For example, the between-group differences in median FT4 and FT3 (0.3 and 0.1 pmol/L respectively) were less than 2% of the median values, whereas between-subject biological variation for these parameters approximates 8-10% (50). Such small differences are unlikely to be relevant to brain development or function, or to autism pathogenesis.

There are limited previous studies of thyroid function in autism for comparison. In a recent case-control study from China of children with neurological disorders, including 1104 ASD cases and 4801 controls (17), FT4 was significantly lower in autistic children, consistent with our findings. In contrast to our results, however, FT3 was significantly higher in cases (the opposite of our result) and TSH significantly lower (whereas we found no significant difference) and total T4 and T3 not significantly different (whereas we found lower levels of each). In each case, the magnitude of the difference was small (0.6 pmol/L for FT4, 0.2 pmol/L for FT3 and 0.13 mU/L for TSH) and seems unlikely to be clinically relevant. The differences between these results and ours are unexplained; they could reflect differences in participant recruitment (including controls), environmental factors, or methodological differences (for example, medication effects were not accounted for in the previous study).

The mechanism underlying the observed differences in thyroid hormones and metabolites in the present study is uncertain. Circulating concentrations of TSH, T4 and T3 are largely heritable traits, so differences could arise from genetic variants, or from environmental or epigenetic influences (18, 51–53). Individual pituitary-thyroid axis setpoints are thought to be established in utero, and can be affected by the intrauterine environment (18, 54) which is also implicated in autism pathogenesis (1). It is possible that the (poorly understood) pathophysiological mechanisms which result in autism also affect hypothalamic development, resulting in altered pituitary-thyroid axis function. Regarding environmental factors, a higher proportion of ASD than non-ASD were taking prescribed medications including anticonvulsants, behavior-modifying and psychotropic drugs. Although we excluded participants on medications with recognized effects on thyroid function tests, subtle effects of other drugs on thyroid hormones and metabolites remain possible. In addition, autistic children often have restricted diets (55), which might alter thyroid hormone metabolism. Mechanistically, a reduction in T4 secretion without alteration in TSH secretion or iodothyronine deiodinase activity could account for the lower circulating levels of the downstream metabolites 3,3’-T2 and 3,5-T2, although it would not account for the higher T0 concentrations.

For the present study, we established customized assays for thyroid hormones and metabolites by LCMS/MS, allowing measurement of total T4, T3, reverse T3, 3,3’-T2, 3,5-T2 and T0 in participants. Assays were also established which detected 7 other thyroid hormone metabolites on spiked sera but were below the limit of detection in participants. There is a limited literature on this methodology in humans for comparison. In a previous study, Jongejan et al. used LCMS/MS to establish a panel of 9 thyroid hormones and metabolites (56). Consistent with our results, they reported undetectable levels of T3A in human sera, but also reported detectable levels of 3-T1 and TA4 which in the present study were below the level of detection. This may reflect differences in LCMS/MS methodology affecting analyte recovery and detection, or possibly developmental effects, since their study was of adults, whereas ours was of children.

Strengths of this study include the large sample size of autistic children, the detailed characterization and expert assessment of participants and the detailed assessment of pituitary-thyroid axis function, including total and free thyroid hormones as well as metabolites by the reference method of LCMS/MS. Our study also has limitations. Timing of blood sampling was not standardized, so results may have been affected by diurnal variation (57) and food intake. We adjusted for age and sex, but not for other factors which are potentially relevant to thyroid function and autism pathogenesis, such as maternal pregnancy history, genetic factors, dietary factors and environmental exposures including endocrine disruptors (58). Arguably, a further limitation is the non-uniform nature of the non-ASD group, which comprised non-autistic siblings of participants and unrelated controls, although this approach has been considered acceptable in previous publications (59–61). In the sensitivity analysis comparing ASD cases with non-autistic sibling and unrelated controls, results were similar to the primary analysis, but it remains possible that comparison with a larger group of unrelated controls would give different results. The only metabolite which differed significantly between non-autistic siblings and unrelated controls was the deiodinated metabolite T0. The basis for this is uncertain, although T0 is thought to be biologically inactive (8).

In conclusion, we report the first study to undertake detailed characterization of pituitary-thyroid axis function and metabolism, including thyroid hormones and metabolites, in a large cohort of autistic children. We found statistically significant differences in circulating thyroid hormones and metabolites in ASD compared with non-ASD, although the magnitude of the differences was small. The study demonstrates the utility of LCMS/MS in detailed characterization of thyroid hormone metabolism, with potential applications in a wide range of clinical and research settings where altered thyroid hormone economy may be present. Further research in independent cohorts would be of interest to confirm our results, to address the discrepant findings between the present study and that of Meng et al. (17). and explore potential mechanisms and implications of altered thyroid hormone economy in autism.

Data availability statement

Thyroid hormone and metabolite data used in this study will be made available by the authors upon request to the Autism CRC/Australian Autism Biobank. Requests to access the datasets should be directed to YmlvYmFua0BhdXRpc21jcmMuY29tLmF1.

Ethics statement

The studies involving humans were approved by University of Western Australia, Human Research Ethics Committee. The studies were conducted in accordance with the local legislation and institutional requirements. Written informed consent for participation in this study was provided by the participants’ legal guardians/next of kin.

Author contributions

MH: Formal analysis, Methodology, Project administration, Writing – original draft, Writing – review & editing, Conceptualization, Data curation, Visualization. RZ: Conceptualization, Data curation, Investigation, Methodology, Project administration, Supervision, Validation, Writing – original draft, Writing – review & editing, Resources, Software, Visualization. SB: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Software, Writing – original draft, Writing – review & editing, Project administration, Resources, Supervision, Validation, Visualization. CB: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Software, Writing – original draft, Writing – review & editing, Resources, Validation. SM: Conceptualization, Data curation, Investigation, Methodology, Writing – original draft, Writing – review & editing, Formal analysis. PC: Conceptualization, Data curation, Investigation, Methodology, Writing – original draft, Writing – review & editing, Formal analysis. SW: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing. EL: Conceptualization, Data curation, Investigation, Methodology, Project administration, Supervision, Validation, Writing – original draft, Writing – review & editing, Formal analysis, Resources, Visualization. AW: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing. JW: Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Resources, Software, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing.

Funding

The author(s) declared that financial support was received for this work and/or its publication. Stan Perron Charitable Foundation Research Grant, 202003067 to JW. Sir Charles Gairdner and Osborne Park Health Care Group Research Grant 2020-21/011 to JW. This research has been conducted using the Australian Autism Biobank, an initiative of Autism CRC, which receives funding from the Australian Government.

Acknowledgments

This research has been conducted using the Australian Autism Biobank, an initiative of Autism CRC. Some of the data used in this project were provided by Autism CRC, with appropriate ethics approval. The Cooperative Research Centre for Living with Autism (Autism CRC) was established and supported under the Australian Government’s Cooperative Research Centre Program. We would like to acknowledge the Australian Autism Biobank participants and their families, as well as The Australian Autism Biobank Team for cohort coordination and data collection: Gail Alvares, Jolene Berry, Vandhana Bharti, Dominique Cleary, Paul Dawson, Cheryl Dissanayake, Valsamma Eapen, Mira Frenk, Jacob Gratten, Rachel Grove, Maryam Haghiran, Alexis Harun, Nicole Hayes, Anjali Henders, Honey Heussler, Helen Holdsworth, Lauren Hollier, Anna Hunt, Rachel Jellett, Feroza Khan, Jodie Leslie, Anne Masi, Tiana McLaren, Candice Michael, Melanie Muniandy, Melissa Neylan, Michaela Nothard, Natalie Silove, Leanne Wallace, Andrew Whitehouse, Naomi Wray.

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fendo.2025.1716586/full#supplementary-material

References

1. Hirota T and King BH. Autism spectrum disorder: A review. Jama. (2023) 329:157–68. doi: 10.1001/jama.2022.23661

PubMed Abstract | Crossref Full Text | Google Scholar

2. Chiarotti F and Venerosi A. Epidemiology of autism spectrum disorders: A review of worldwide prevalence estimates since 2014. Brain Sci. (2020) 10:274–94. doi: 10.3390/brainsci10050274

PubMed Abstract | Crossref Full Text | Google Scholar

3. Li Z, Yang L, Chen H, Fang Y, Zhang T, Yin X, et al. Global, regional and national burden of autism spectrum disorder from 1990 to 2019: results from the Global Burden of Disease Study 2019. Epidemiol Psychiatr Sci. (2022) 31:e33. doi: 10.1017/s2045796022000178

PubMed Abstract | Crossref Full Text | Google Scholar

4. Wei H, Zhu Y, Wang T, Zhang X, Zhang K, and Zhang Z. Genetic risk factors for autism-spectrum disorders: a systematic review based on systematic reviews and meta-analysis. J Neural Transm. (2021) 128:717–34. doi: 10.1007/s00702-021-02360-w

PubMed Abstract | Crossref Full Text | Google Scholar

5. Modabbernia A, Velthorst E, and Reichenberg A. Environmental risk factors for autism: an evidence-based review of systematic reviews and meta-analyses. Mol Autism. (2017) 8:13–28. doi: 10.1186/s13229-017-0121-4

PubMed Abstract | Crossref Full Text | Google Scholar

6. Salas-Lucia F. Mapping thyroid hormone action in the human brain. Thyroid. (2024) 34:815–26. doi: 10.1089/thy.2024.0120

PubMed Abstract | Crossref Full Text | Google Scholar

7. Bernal J. Thyroid hormones in brain development and function. In: Feingold KR, Ahmed SF, Anawalt B, Blackman MR, Boyce A, Chrousos G, et al, editors. Endotext. MDText.com, Inc, South Dartmouth (MA (2000).

Google Scholar

8. Köhrle J. The colorful diversity of thyroid hormone metabolites. Eur Thyroid J. (2019) 8:115–29. doi: 10.1159/000497141

PubMed Abstract | Crossref Full Text | Google Scholar

9. Richards KH, Schanze N, Monk R, Rijntjes E, Rathmann D, and Köhrle J. A validated LC-MS/MS method for cellular thyroid hormone metabolism: Uptake and turnover of mono-iodinated thyroid hormone metabolites by PCCL3 thyrocytes. PloS One. (2017) 12:e0183482. doi: 10.1371/journal.pone.0183482

PubMed Abstract | Crossref Full Text | Google Scholar

10. Samuels MH. Thyroid disease and cognition. Endocrinol Metab Clin North Am. (2014) 43:529–43. doi: 10.1016/j.ecl.2014.02.006

PubMed Abstract | Crossref Full Text | Google Scholar

11. Tost M, Monreal JA, Armario A, Barbero JD, Cobo J, García-Rizo C, et al. Targeting hormones for improving cognition in major mood disorders and schizophrenia: thyroid hormones and prolactin. Clin Drug Investig. (2020) 40:1–14. doi: 10.1007/s40261-019-00854-w

PubMed Abstract | Crossref Full Text | Google Scholar

12. Noda M. Possible role of glial cells in the relationship between thyroid dysfunction and mental disorders. Front Cell Neurosci. (2015) 9:194. doi: 10.3389/fncel.2015.00194

PubMed Abstract | Crossref Full Text | Google Scholar

13. Shrestha S, Bloom MS, Yucel R, Seegal RF, Rej R, McCaffrey RJ, et al. Thyroid function and neuropsychological status in older adults. Physiol Behav. (2016) 164:34–9. doi: 10.1016/j.physbeh.2016.05.037

PubMed Abstract | Crossref Full Text | Google Scholar

14. Medici M, Direk N, Visser WE, Korevaar TI, Hofman A, Visser TJ, et al. Thyroid function within the normal range and the risk of depression: a population-based cohort study. J Clin Endocrinol Metab. (2014) 99:1213–9. doi: 10.1210/jc.2013-3589

PubMed Abstract | Crossref Full Text | Google Scholar

15. Andersen SL, Andersen S, Vestergaard P, and Olsen J. Maternal thyroid function in early pregnancy and child neurodevelopmental disorders: A danish nationwide case-cohort study. Thyroid. (2018) 28:537–46. doi: 10.1089/thy.2017.0425

PubMed Abstract | Crossref Full Text | Google Scholar

16. Alvarez-Pedrerol M, Ribas-Fitó N, Torrent M, Julvez J, Ferrer C, and Sunyer J. TSH concentration within the normal range is associated with cognitive function and ADHD symptoms in healthy preschoolers. Clin Endocrinol. (2007) 66:890–8. doi: 10.1111/j.1365-2265.2007.02871.x

PubMed Abstract | Crossref Full Text | Google Scholar

17. Meng H, Bigambo FM, Gu W, Wang X, and Li Y. Evaluation of thyroid function tests among children with neurological disorders. Front Endocrinol. (2024) 15:1498788. doi: 10.3389/fendo.2024.1498788

PubMed Abstract | Crossref Full Text | Google Scholar

18. Walsh JP. Thyroid function across the lifespan: do age-related changes matter? Endocrinol Metab. (2022) 37:208–19. doi: 10.3803/EnM.2022.1463

PubMed Abstract | Crossref Full Text | Google Scholar

19. Chen Y, Luo ZC, Zhang T, Fan P, Ma R, Zhang J, et al. Maternal thyroid dysfunction and neuropsychological development in children. J Clin Endocrinol Metab. (2023) 108:339–50. doi: 10.1210/clinem/dgac577

PubMed Abstract | Crossref Full Text | Google Scholar

20. Korevaar TI, Muetzel R, Medici M, Chaker L, Jaddoe VW, de Rijke YB, et al. Association of maternal thyroid function during early pregnancy with offspring IQ and brain morphology in childhood: a population-based prospective cohort study. Lancet Diabetes Endocrinol. (2016) 4:35–43. doi: 10.1016/s2213-8587(15)00327-7

PubMed Abstract | Crossref Full Text | Google Scholar

21. Kahn A. Thyroid dysfunction. Br Med J. (1970) 4:495. doi: 10.1136/bmj.4.5733.495-d

PubMed Abstract | Crossref Full Text | Google Scholar

22. Campbell M, Small AM, Hollander CS, Korein J, Cohen IL, Kalmijn M, et al. A controlled crossover study of triiodothyronine in autistic children. J Autism Child Schizophr. (1978) 8:371–81. doi: 10.1007/bf01538042

PubMed Abstract | Crossref Full Text | Google Scholar

23. Khan A, Harney JW, Zavacki AM, and Sajdel-Sulkowska EM. Disrupted brain thyroid hormone homeostasis and altered thyroid hormone-dependent brain gene expression in autism spectrum disorders. J Physiol Pharmacol. (2014) 65:257–72.

PubMed Abstract | Google Scholar

24. Abbassi V, Linscheid T, and Coleman M. Triiodothyronine (T3) concentration and therapy in autistic children. J Autism Child Schizophr. (1978) 8:383–7. doi: 10.1007/bf01538043

PubMed Abstract | Crossref Full Text | Google Scholar

25. Cohen DJ, Young JG, Lowe TL, and Harcherik D. Thyroid hormone in autistic children. J Autism Dev Disord. (1980) 10:445–50. doi: 10.1007/bf02414820

PubMed Abstract | Crossref Full Text | Google Scholar

26. Wu S, Ding Y, Wu F, Li R, Xie G, Hou J, et al. Family history of autoimmune diseases is associated with an increased risk of autism in children: A systematic review and meta-analysis. Neurosci Biobehav Rev. (2015) 55:322–32. doi: 10.1016/j.neubiorev.2015.05.004

PubMed Abstract | Crossref Full Text | Google Scholar

27. Román GC, Ghassabian A, Bongers-Schokking JJ, Jaddoe VW, Hofman A, de Rijke YB, et al. Association of gestational maternal hypothyroxinemia and increased autism risk. Ann Neurol. (2013) 74:733–42. doi: 10.1002/ana.23976

PubMed Abstract | Crossref Full Text | Google Scholar

28. Kaplan ZB, Pearce EN, Lee SY, Shin HM, and Schmidt RJ. Maternal thyroid dysfunction during pregnancy as an etiologic factor in autism spectrum disorder: challenges and opportunities for research. Thyroid. (2024) 34:144–57. doi: 10.1089/thy.2023.0391

PubMed Abstract | Crossref Full Text | Google Scholar

29. Faruk MO, Rahman MS, Rana MS, Mahmud S, Al-Neyma M, Karim MS, et al. Socioeconomic and demographic risk factors of autism spectrum disorder among children and adolescents in Bangladesh: Evidence from a cross-sectional study in 2022. PloS One. (2023) 18:e0289220. doi: 10.1371/journal.pone.0289220

PubMed Abstract | Crossref Full Text | Google Scholar

30. Ge GM, Leung MT, Man KK, Leung WC, Ip P, Li GH, et al. Maternal thyroid dysfunction during pregnancy and the risk of adverse outcomes in the offspring: A systematic review and meta-analysis. J Clin Endocrinol Metab. (2020) 105:3821–41. doi: 10.1210/clinem/dgaa555

PubMed Abstract | Crossref Full Text | Google Scholar

31. Andersen SL, Laurberg P, Wu CS, and Olsen J. Attention deficit hyperactivity disorder and autism spectrum disorder in children born to mothers with thyroid dysfunction: a Danish nationwide cohort study. Bjog. (2014) 121:1365–74. doi: 10.1111/1471-0528.12681

PubMed Abstract | Crossref Full Text | Google Scholar

32. Getahun D, Jacobsen SJ, Fassett MJ, Wing DA, Xiang AH, Chiu VY, et al. Association between maternal hypothyroidism and autism spectrum disorders in children. Pediatr Res. (2018) 83:580–8. doi: 10.1038/pr.2017.308

PubMed Abstract | Crossref Full Text | Google Scholar

33. Rotem RS, Chodick G, Shalev V, Davidovitch M, Koren G, Hauser R, et al. Maternal thyroid disorders and risk of autism spectrum disorder in progeny. Epidemiology. (2020) 31:409–17. doi: 10.1097/ede.0000000000001174

PubMed Abstract | Crossref Full Text | Google Scholar

34. González-Madrid E, Rangel-Ramírez MA, Opazo MC, Méndez L, Bohmwald K, Bueno SM, et al. Gestational hypothyroxinemia induces ASD-like phenotypes in behavior, proinflammatory markers, and glutamatergic protein expression in mouse offspring of both sexes. Front Endocrinol. (2024) 15:1381180. doi: 10.3389/fendo.2024.1381180

PubMed Abstract | Crossref Full Text | Google Scholar

35. Alvares GA, Dawson PA, Dissanayake C, Eapen V, Gratten J, Grove R, et al. Study protocol for the Australian autism biobank: an international resource to advance autism discovery research. BMC Pediatr. (2018) 18:284–92. doi: 10.1186/s12887-018-1255-z

PubMed Abstract | Crossref Full Text | Google Scholar

36. Lord C, Rutter M, DiLavore P, Risi S, Gotham K, and Bishop S. Autism diagnostic observation schedule: ADOS. 2nd ed. Torrance, CA, USA: Western Psychological Services (2012).

Google Scholar

37. Gotham K, Pickles A, and Lord C. Standardizing ADOS scores for a measure of severity in autism spectrum disorders. J Autism Dev Disord. (2009) 39:693–705. doi: 10.1007/s10803-008-0674-3

PubMed Abstract | Crossref Full Text | Google Scholar

38. Mullen E. Mullen scales of early learning, vol. AGS. Circle Pines, MN: American Guidance Service Inc (1995).

Google Scholar

39. Wechsler D. Wechsler abbreviated scale of intelligence, vol. 2. Pearson: San Antonio TX (2011).

Google Scholar

40. Sparrow S, Cicchettim D, and Balla D. Vineland adaptive behavior scales. 2nd ed. Circle Pines, MN, USA: Vineland-II): APA PsycTests (2005).

Google Scholar

41. Botelho J, Bachmann L, Cao Z, Chandler D, Clarke N, Demers L, et al. Mass spectrometry for androgen and estrogen measurements in serum. P35 Wayne, PA, USA: Clinical and Laboratory Standards Institute (2015).

Google Scholar

42. Jostel A, Ryder WD, and Shalet SM. The use of thyroid function tests in the diagnosis of hypopituitarism: definition and evaluation of the TSH Index. Clin Endocrinol. (2009) 71:529–34. doi: 10.1111/j.1365-2265.2009.03534.x

PubMed Abstract | Crossref Full Text | Google Scholar

43. Sun H, Zhu W, Liu J, An Y, Wang Y, and Wang G. Reduced sensitivity to thyroid hormones is associated with high remnant cholesterol levels in chinese euthyroid adults. J Clin Endocrinol Metab. (2022) 108:166–74. doi: 10.1210/clinem/dgac523

PubMed Abstract | Crossref Full Text | Google Scholar

44. R_Core_Team. R: A language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing (2013). Available online at: http://www.R-project.org/. (Accessed December 01, 2023)

Google Scholar

45. Delange F. Neonatal hypothyroidism: recent developments. Baillieres Clin Endocrinol Metab. (1988) 2:637–52. doi: 10.1016/s0950-351x(88)80057-0

PubMed Abstract | Crossref Full Text | Google Scholar

46. Haddow JE, Palomaki GE, Allan WC, Williams JR, Knight GJ, Gagnon J, et al. Maternal thyroid deficiency during pregnancy and subsequent neuropsychological development of the child. N Engl J Med. (1999) 341:549–55. doi: 10.1056/nejm199908193410801

PubMed Abstract | Crossref Full Text | Google Scholar

47. Vanderpump MP. The epidemiology of thyroid disease. Br Med Bull. (2011) 99:39–51. doi: 10.1093/bmb/ldr030

PubMed Abstract | Crossref Full Text | Google Scholar

48. Senese R, De Lange P, Petito G, Moreno M, Goglia F, and Lanni A. 3,5-diiodothyronine: A novel thyroid hormone metabolite and potent modulator of energy metabolism. Front Endocrinol. (2018) 9:427. doi: 10.3389/fendo.2018.00427

PubMed Abstract | Crossref Full Text | Google Scholar

49. Olvera A, Martyniuk CJ, Buisine N, Jiménez-Jacinto V, Sanchez-Flores A, Sachs LM, et al. Differential transcriptome regulation by 3,5-T2 and 3’,3,5-T3 in brain and liver uncovers novel roles for thyroid hormones in tilapia. Sci Rep. (2017) 7:15043. doi: 10.1038/s41598-017-14913-9

PubMed Abstract | Crossref Full Text | Google Scholar

50. Mairesse A, Wauthier L, Courcelles L, Luyten U, Burlacu MC, Maisin D, et al. Biological variation and analytical goals of four thyroid function biomarkers in healthy European volunteers. Clin Endocrinol. (2021) 94:845–50. doi: 10.1111/cen.14356

PubMed Abstract | Crossref Full Text | Google Scholar

51. Mulder TA, Campbell PJ, Taylor PN, Peeters RP, Wilson SG, Medici M, et al. Genetic determinants of thyroid function in children. Eur J Endocrinology. (2023) 189:164–74. doi: 10.1093/ejendo/lvad086

PubMed Abstract | Crossref Full Text | Google Scholar

52. Weihs A, Chaker L, Martin TC, Braun KV, Campbell PJ, Cox SR, et al. Epigenome-wide association study reveals cpG sites associated with thyroid function and regulatory effects on KLF9. Thyroid. (2023) 33:301–11. doi: 10.1089/thy.2022.0373

PubMed Abstract | Crossref Full Text | Google Scholar

53. Sterenborg R, Steinbrenner I, Li Y, Bujnis MN, Naito T, Marouli E, et al. Multi-trait analysis characterizes the genetics of thyroid function and identifies causal associations with clinical implications. Nat Commun. (2024) 15:888–906. doi: 10.1038/s41467-024-44701-9

PubMed Abstract | Crossref Full Text | Google Scholar

54. Päkkilä F, Männistö T, Surcel HM, Ruokonen A, Bloigu A, Pouta A, et al. Maternal thyroid dysfunction during pregnancy and thyroid function of her child in adolescence. J Clin Endocrinol Metab. (2013) 98:965–72. doi: 10.1210/jc.2012-2028

PubMed Abstract | Crossref Full Text | Google Scholar

55. Mathew NE, Mallitt KA, Masi A, Katz T, Walker AK, Morris MJ, et al. Dietary intake in children on the autism spectrum is altered and linked to differences in autistic traits and sensory processing styles. Autism Res. (2022) 15:1824–39. doi: 10.1002/aur.2798

PubMed Abstract | Crossref Full Text | Google Scholar

56. Jongejan RM, Klein T, Meima ME, Visser WE, van Heerebeek RE, Luider TM, et al. A mass spectrometry-based panel of nine thyroid hormone metabolites in human serum. Clin Chem. (2020) 66:556–66. doi: 10.1093/clinchem/hvaa022

PubMed Abstract | Crossref Full Text | Google Scholar

57. Zhang Y, Wang HL, Zhou CQ, He DH, Wu F, Li HC, et al. Intraday changes and clinical applications of thyroid function biomarkers in healthy subjects. Lab Med. (2023) 54:282–6. doi: 10.1093/labmed/lmac105

PubMed Abstract | Crossref Full Text | Google Scholar

58. He J, Xu J, Zheng M, Pan K, Yang L, Ma L, et al. Thyroid dysfunction caused by exposure to environmental endocrine disruptors and the underlying mechanism: A review. Chem Biol Interact. (2024) 391:110909. doi: 10.1016/j.cbi.2024.110909

PubMed Abstract | Crossref Full Text | Google Scholar

59. Yap CX, Henders AK, Alvares GA, Giles C, Huynh K, Nguyen A, et al. Interactions between the lipidome and genetic and environmental factors in autism. Nat Med. (2023) 29:936–49. doi: 10.1038/s41591-023-02271-1

PubMed Abstract | Crossref Full Text | Google Scholar

60. Lin PI, John JR, Masi A, Ong LK, Mathew NE, Moni MA, et al. The relationship between hair cortisol concentration and autism diagnosis. J Psychiatr Res. (2024) 176:68–76. doi: 10.1016/j.jpsychires.2024.05.052

PubMed Abstract | Crossref Full Text | Google Scholar

61. Mullin BH, Stuckey M, Brown SJ, Mullin S, Campbell PJ, Dudbridge F, et al. Plasma metabolite profiles of children with autism spectrum disorder. Metabolites. (2025) 15:780. doi: 10.3390/metabo15120780

PubMed Abstract | Crossref Full Text | Google Scholar

62. Bohn MK, Higgins V, Asgari S, Leung F, Hoffman B, Macri J, et al. Paediatric reference intervals for 17 Roche cobas 8000 e602 immunoassays in the CALIPER cohort of healthy children and adolescents. Clin Chem Lab Med. (2019) 57:1968–79. doi: 10.1515/cclm-2019-0707

PubMed Abstract | Crossref Full Text | Google Scholar