The Causes & Treatments for Autism | Dr. Karen Parker

Current US prevalence is about 1 in 36 children, up from 1 in 44 just a few years ago, with a 3–4:1 male bias. Pediatricians now use early screening tools and can often diagnose by age 2–3, yet specialist waitlists can run 12–18 months. Because earlier behavioral interventions are more effective, there is urgent need for scalable lab or tech-based pre‑screening tools (e.g., biomarkers, eye‑gaze tests) to prioritize high‑risk children for evaluation.

Autism is defined behaviorally by two core domains—pervasive social interaction challenges and restricted, repetitive behaviors—but presentations vary dramatically. Many individuals also have anxiety, sensory issues, seizures, or sleep problems. Genetic data suggest both highly penetrant single‑gene syndromes (e.g., Fragile X, Timothy syndrome) and a broad polygenic background. Without clear biological subtypes, trials lump very different patients together, diluting treatment signals.

Large CSF and blood studies show oxytocin levels do not systematically differ between autistic and non‑autistic groups. In a small four‑week intranasal oxytocin trial, Parker found that only children with low baseline blood oxytocin showed meaningful social improvement; those with normal or high levels did not. A later large multi‑site oxytocin trial found no overall benefit, and technical issues in peptide handling likely further obscured any subgroup effects. Oxytocin may still help young children (e.g., ages ~2–6) or low‑oxytocin subgroups, but funding for such targeted trials has largely evaporated.

Using a naturalistic rhesus macaque model of low vs. high sociability, Parker’s lab showed that lower CSF vasopressin (but not oxytocin) strongly predicted reduced grooming and social engagement, correctly classifying monkeys as high vs. low social with ~93% accuracy. She then piggybacked on clinically indicated lumbar punctures in children and replicated the finding: CSF vasopressin levels were markedly lower in autistic vs. non‑autistic children, and lower vasopressin correlated with greater social symptom severity on gold‑standard diagnostic instruments.

In collaboration with John Constantino, Parker analyzed banked neonatal CSF samples from medically evaluated but largely healthy infants. Retrospective chart review identified which infants later received autism diagnoses. Those future‑autism infants had significantly lower CSF vasopressin at infancy than those who did not develop autism, while CSF oxytocin did not differ. This supports the idea that low vasopressin is present before behavioral symptoms emerge and may be part of the causal pathway, not just a consequence of social difficulties.