The research team, led by Feng-Yun Zou, Jia-Yuan Zhao, Shuang-Shuang Zhong and Bo-Ya Yin, recruited 460 children for the study. The cohort comprised 140 children with autism spectrum disorder who also showed elevated anxiety symptoms, 160 children with autism but without elevated anxiety, and 160 typically developing controls. This three-group design is what gives the study its power: rather than simply comparing autistic children with neurotypical peers, the investigators could isolate the features that specifically track anxiety within the autism population, a distinction that most earlier imaging studies blurred.

The imaging method at the heart of the study is called mean apparent propagator MRI, or MAP-MRI, and it represents a significant technical step beyond conventional diffusion-weighted imaging. Standard diffusion MRI measures the average direction and extent of water movement in tissue, which is useful for detecting gross white matter changes but says little about the complexity of the cellular environment. MAP-MRI instead reconstructs a full three-dimensional probability distribution of water molecule displacements, allowing researchers to quantify how restricted, hindered or free the diffusion is within each voxel. From this propagator, the team derived eight metrics for each of 68 brain regions, capturing properties such as non-Gaussianity of diffusion, return-to-origin probability and the orientation coherence of tissue microstructure.

Why does this matter biologically? Non-Gaussian diffusion measures are sensitive to the density and packing of cells, the tortuosity of the extracellular space and the organization of neurites, all of which can be altered by neuroinflammation, atypical development or altered neuronal arborization. In effect, MAP-MRI gives researchers a microscope-like window into tissue architecture without a biopsy. The regions that emerged as most different in the anxious autism group were exactly those one would predict from the neuroscience of fear and emotion: the amygdala, the hippocampus, the insula, the rostral anterior cingulate cortex, the putamen and the globus pallidus. The largest and most consistent differences were in non-Gaussian diffusion effects across this distributed limbic and striatal network.

The amygdala and hippocampus are the canonical structures of threat processing and memory for fearful experiences, while the insula integrates bodily sensations with emotional awareness and the anterior cingulate governs conflict monitoring and emotional regulation. The putamen and globus pallidus, core components of the striatal system, are increasingly recognized as participants in repetitive behavior and reward-related learning, both of which intersect with anxiety in autism. Finding that microstructural differences in these regions specifically distinguish anxious autistic children from their non-anxious autistic peers suggests that anxiety in autism is not merely a psychological overlay but is reflected in measurable tissue-level organization of the emotional brain.

Alongside the imaging, the researchers assayed a panel of peripheral biomarkers from fasting venous blood: free triiodothyronine (FT3), total thyroxine (TT4), thyroid-stimulating hormone (TSH), serum iron, blood lead and magnesium. The choice was deliberate. Thyroid hormones are master regulators of brain development and neuronal excitability, iron is essential for myelination and neurotransmitter synthesis, and lead is a well-established neurotoxicant with no safe exposure threshold. Magnesium, meanwhile, plays roles in neural excitability and stress responses. The results showed that children with autism and elevated anxiety had higher levels of FT3, TT4 and blood lead, and lower serum iron, compared with both non-anxious autistic children and typically developing controls.

The correlation analyses within the anxious autism subgroup revealed a coherent pattern. Serum iron was positively associated with MAP-MRI metrics, meaning that higher iron levels accompanied microstructural features consistent with more organized tissue in the affected regions. Blood lead and thyroid hormones showed the opposite relationship, with inverse associations to the same imaging measures, most prominently in the amygdala and hippocampus. In other words, the two metals and the thyroid hormones appeared to sit on opposite sides of the same microstructural signature, with the strongest signal concentrated in the brain’s fear circuitry.

To probe how these biochemical measures might relate to one another, the team applied exploratory cross-sectional statistical association decomposition models, which partitioned the observed correlations into direct components and indirect components mediated through FT3. This kind of decomposition cannot prove causation, but it provides a structured way of asking whether thyroid hormone status might act as a pathway linking metal exposure to brain microstructure. The models suggested that part of the association between the peripheral measures and the imaging features flowed through FT3-related indirect pathways, hinting at a possible endocrine intermediary role, though the authors are careful to frame this as hypothesis-generating rather than confirmatory.

Adding an unusual cross-species dimension, the researchers included a small, independent exploratory component using BTBR mice, an inbred strain commonly used as a model for autism-like behaviors because the animals display reduced social interaction, repetitive self-grooming and anxiety-like responses. In mice showing anxiety-like behavior, the team observed reduced amygdala fiber area and disrupted fiber organization, a pattern that echoes the human imaging findings in the same brain region. While the animal work is explicitly labeled exploratory and involves a small sample, the convergence between the mouse fiber architecture data and the human MAP-MRI abnormalities in the amygdala lends a degree of biological plausibility to the overall picture.

The authors and outside observers alike stress the limits of the design. All measurements were taken at a single point in time, so the study cannot determine whether elevated thyroid hormones and lead precede the microstructural differences, result from them, or both reflect some third factor such as chronic stress or shared genetic vulnerability. The children were not followed longitudinally, and no intervention was tested. What the study does establish is a detailed, multiplicity-corrected association profile: children with autism and elevated anxiety carry a distinctive combination of higher FT3, TT4 and blood lead, lower iron, and distributed MAP-MRI abnormalities centered on the limbic and striatal circuits that generate and regulate fear. If future longitudinal and experimental studies confirm these pathways, the findings could eventually point toward biomarker-guided subtyping of anxiety within autism, opening the door to targeted screening for thyroid function and metal exposure in children whose anxiety symptoms are hardest to treat. For now, the study stands as a compelling demonstration that the heterogeneity of autism is written not only in behavior but in the microscopic fabric of the brain and the chemistry of the blood that nourishes it.

Subject of Research: Associations between thyroid hormones, trace elements and brain microstructure in anxiety-related heterogeneity among children with autism spectrum disorder

Article Title: Mean apparent propagator MRI of distributed brain microstructure in anxiety-related heterogeneity among children with autism spectrum disorder: associations with thyroid hormones and trace elements

Article References: Zou, F.-Y., Zhao, J.-Y., Zhong, S.-S., Yin, B.-Y., Ni, Y.-X., Deng, Y.-Y., Zhou, X., Luo, X.-W., Shen, L.-S., Zhang, J.-L., Lin, Z.-P., Zhou, W.-Y., Wang, S.-H., & Guo, R.-M. (2026). Mean apparent propagator MRI of distributed brain microstructure in anxiety-related heterogeneity among children with autism spectrum disorder: associations with thyroid hormones and trace elements. BMC Psychiatry. https://doi.org/10.1186/s12888-026-08637-2

Image Credits: AI Generated

DOI: 10.1186/s12888-026-08637-2

Keywords: autism spectrum disorder, anxiety, MAP-MRI, diffusion MRI, amygdala, hippocampus, thyroid hormones, trace elements, blood lead, iron, neuroimaging, pediatric psychiatry