Precision medicine, an approach to treating disease that accounts for individual variability in genes, environment, and lifestyle, necessitates gathering diverse data. Rather than trying to gather as much data as possible, the intersections of different kinds of data are thought to enable identification of different subgroups of patients that have different combinations of disease and individual traits. In this study, the researchers test this thought using autism. The researchers combined healthcare claims, electronic health records, familial exome sequences, and gene expression patterns to identify a subgroup of patients that have autism associated with dyslipidemia (abnormal amount of lipids in the blood).
The researchers wanted to test the thought that overlapping different kinds of data could identify patient subgroups that could be targeted by precision medicine. They used autism to test this thought.
The National Academy of Sciences published a Precision Medicine Report that said if multiple molecular indicators of disease are used along other kidneys of data, they can more accurately describe and classify diseases and their treatment. The researchers tested whether this approach could identify disease subtypes using autism spectrum disorder (ASD), which is estimated to affect 1 in 54 children in the U.S. Currently, ASD has no effective treatment and has clinical and genetic heterogeneity, highlighting the need for identifying patient subgroups to improve diagnosis and outcomes.
The researchers integrated large datasets of familial exome sequences, genetic expression patterns, electronic health records, and healthcare claims. They looked at clusters of different disease variations to identify variants of ASD that were of interest. First, they retrieved data of genetic expression in the brain during development from 524 samples from 26 brain regions of 42 people. ASD is thought to be driven by different gene variants during brain development, so the researchers identified clusters of genes that are expressed together during early brain development. They especially focused on clusters that are expressed differently in males and females, since males are 4 times as likely to have ASD than females.
Next, the researchers also compiled familial exome sequences of 3531 individuals from 1704 families with 1 child who has ASD and 1 child who doesn’t (simplex families), as well as 50 families with 2-5 affected siblings (multiplex families). They identified different variants between siblings for simplex families, and variants that are similar between siblings for multiplex families. They focused on heritable mutations that could disrupt genes expressed during brain development.
After gathering these variants, the researchers mapped them with the clusters of genes to identify co-expressed genes that are involved in ASD and have sex-specific effects during brain development. They used affected sibling-pair analysis to assess the significance of variant sharing in multiplex families, and permutation tests in singleplex families to assess variant differences between affected siblings and unaffected siblings. This analysis identified 33 co-expressed, sex-differentiated clusters of genes that were associated with ASD. Figure 1 shows how different sources of data were used to create associations between genes and create clusters differentially expressed by sex and that were associated with ASD.
Analysis of these 33 clusters showed that they shared several molecular patterns previously described in ASD, but that they also identified an unrecognized molecular pattern for lipid regulation. ASD-associated gene variants predicted to collectively alter low-density lipoproteins, cholesterol, and triglyceride levels. The researchers tested the hypothesis that dyslipidemia might be associated with ASD by comparing bloog lipid profiles and dyslipidemia diagnoses in individuals with ASD, as well as in their family members and unrelated individuals (controls). They used the records of 2,750,021 children at Boston Children’s hospital, including 25,514 children with ASD, and compared lipid lab tests between children with ASD and similar children that didn’t have ASD. The researchers found that children with ASD have blood lipid levels significantly outside the average range, even after adjusting for age, sex, and metabolism.
The researchers also assessed coincidences of ASD and dyslipidemia using healthcare claims data from 34,003,107 individuals, with 80,714 diagnosed with ASD. The researchers matched these individuals with similar controls, and found a higher incidence of dyslipidemia in individuals with ASD. They also found that diagnoses of dyslipidemia in both mothers and fathers were associated with ASD in their children. Even after controlling for genetic background and familial eating habits, dyslipidemia and ASD were associated in families. Dyslipidemia was found in 6.55% of individuals with ASD.
Next, the researchers compared ASD-associated characteristics between individuals with ASD who had and didn’t have dyslipidemia. Epilepsy, sleep disorders, and ADHD were associated with individuals who had ASD and dyslipidemia, suggesting that dyslipidemia may be associated with brain development in general. In addition, several endocrine and metabolism-related diagnoses were associated with individuals with ASD and dyslipidemia, including hypothyroidism, anemia, and vitamin D deficiency.
Even after controlling for common prescription drugs that alter lipid levels, dyslipidemia was still associated with ASD in individuals with dyslipidemia and ASD.
Lastly, the researchers compared the phenotypes of engineered model organisms, specifically mice, of dyslipidemia and ASD gene dysfunction. After investigating 1,315 reported phenotypes, the models demonstrated that ASD mouse models have phenotypes more similar to dyslipidemia phenotypes than control mouse models. Conversely, Mouse models of dyslipidemia were found to have social and nervous system abnormalities similar to those in ASD models. Figure 2 shows the associations between ASD and lipid regulation functions.
In this study, the researchers characterized a subtype of ASD associated with dyslipidemia. These findings are consistent with reported functions of dyslipidemia genes in brain development, as well as previous analyses of ASD-associated genes that reported increased expression of lipid-associated genes. In addition, Rett syndrome and Smith-Lemli-Opitz syndrome (SLOS) are both lipid metabolism-related disorders associated with the autism spectrum, which may be explained by the findings of this study. The findings support the identification of disease subtypes that may be targeted for tailored intervention and prevention, such as in oncology.