Tracking autism risk from birth
Autism spectrum disorders (ASDs) affect millions of families worldwide, yet diagnosis often comes only after behavioural symptoms become evident. By that stage, opportunities for earlier support or preventive intervention may already have been missed. Understanding how genetic susceptibility, biology and the environment interact during infancy is therefore a major priority for researchers and healthcare systems.
Following autism risk from birth
The EU-funded GEMMA(opens in new window) project set out to identify early biomarkers of ASD by following infants of elevated familial risk from birth to 36 months of age. The study recruited approximately 350 newborns from families where an older sibling had already been diagnosed with autism. “The project offered a unique opportunity to observe the natural history of autism before clinical diagnosis. By tracking biological and behavioural changes over time, we aimed to identify risk onset and how it might be modified,” explains project coordinator Alessio Fasano. Participants provided biological samples and clinical information every six months, and underwent regular behavioural assessments. This longitudinal design enabled researchers to capture the biological changes occurring before, during and after symptom onset. GEMMA combined multiple layers of information, including genomic, epigenomic, microbiome and metabolomic data. This integrated systems biology strategy was designed to explore how inherited susceptibility interacts with gut bacteria, immune responses and metabolism. Machine learning tools were then applied to detect patterns associated with later autism development(opens in new window). “This allowed the team to move from isolated risk factors towards a predictive model based on the interaction of many biological parameters,” adds Fasano.
Early warning signs
GEMMA discovered that changes in the gut microbiome and its function may precede behavioural symptoms by several months. Researchers identified shifts associated with elevated zonulin(opens in new window), a biomarker linked to gut permeability and inflammatory processes. “This suggests that biological changes can occur well before traditional diagnosis is possible,” highlights Fasano. The study also found that meaningful behavioural indicators may emerge earlier than previously recognised. Infants aged nine months who were described as unusually fussy, difficult to calm or prone to sleep difficulties were more likely to show early autism-related signs at 12 months.
Microbiome-based interventions
GEMMA also explored whether modifying the gut microbiome(opens in new window) could help alleviate autism behaviour. A six-month open-label intervention combining probiotics and prebiotics showed improvements in gastrointestinal symptoms and behavioural challenges. Prospective analyses additionally identified a set of bacterial taxa able to predict later autism development with high accuracy. These findings were supported in humanised animal models using faecal microbiota transfer, helping to strengthen evidence for a causal role of gut microbial imbalance. By integrating cutting-edge data science with real-world clinical follow-up, GEMMA has helped redefine autism research towards prediction, prevention and personalised care. Looking ahead, international collaborations are already under way to validate the findings in additional populations and translate them into practical tools. In the longer term, the project envisions clinical pathways in which infants with a family history of autism could receive integrated risk assessments combining genetics, gut microbiome profiles, immune markers and behavioural observations. Predictive algorithms could then help clinicians identify children most likely to benefit from early personalised interventions, such as targeted nutritional or probiotic strategies, before symptoms fully emerge.