Cerebrospinal fluid metabolomics in autistic regression reveals dysregulation of sphingolipids and decreased β-hydroxybutyrate.

Abstract

Background: Autism is highly heritable, however actionable genetic findings are only found in a minority of patients. Many people with autism suffer loss of neurodevelopmental skills, known as autistic regression. The cause of regression is poorly understood, and the diagnostic and therapeutic pathways are lacking.

Methods: We used untargeted metabolomics using a UPLC-Q-Exactive-HFx Mass Spectrometry to examine cerebrospinal fluid (CSF) from twenty-two patients with autistic regression compared to sixteen controls with neurodevelopmental disorders (but not autistic regression) and thirty-four controls with other neurological disease (headache, encephalitis, epilepsy). The twenty-two patients with autistic regression consisted of two groups: early (infantile) autistic regression <2 years of age (n = 8), and later regression of skills >4 years of age, often in the context of pre-existing developmental concerns (n = 14). Metabolites of interest were then quantified and validated using targeted assays.

Findings: Untargeted case-control studies revealed good separation of patients from controls using multivariate analysis. β-hydroxybutyrate was significantly decreased in the CSF of patients with autistic regression, and the findings were validated using a targeted β-hydroxybutyrate assay. The sphingolipid, sphingosine-1-phosphate was significantly elevated in the discovery case-control studies, and sphingolipid metabolism pathways were also significantly dysregulated. We therefore developed a targeted metabolite assay of forty sphingolipids. After FDR correction, 21 of the 40 sphingolipids were significantly dysregulated (p_FDR < 0.05) (Benjamini-Hochberg correction) in autistic regression compared to the neurodevelopmental controls, and 26 of the 40 sphingolipids were significantly dysregulated in autistic regression compared to other neurological controls, with elevated ceramides, hexosylceramides, sphingosines (including sphingosine-1-phosphate), and sulfatides. By contrast, sphingomyelin levels were generally decreased in autistic regression.

Interpretation: Our data shows the potential utility of CSF metabolomics in the context of autistic regression, a clinical syndrome which has historically lacked pathophysiological biomarkers and disease modifying therapies.

Funding: Financial support for the study was granted by Dale NHMRC Investigator grant APP1193648, Petre Foundation, Cerebral Palsy Alliance, and Ainsworth and SCHF Neuroscience grant scheme.