New study suggests regressive autism is a distinct subtype with unique genetic and clinical features

A recent analysis of children in a Paris autism cohort indicates that regressive autism may be more than a variation within the spectrum , potentially representing a separate clinical and biological entity, prompting a need for targeted early detection and intervention strategies.

A new analysis of children in a Paris autism cohort suggests that regressive autism may be a clinically distinct subgroup rather than simply part of the wider spectrum. In the study, researchers reviewed early development, assessed symptom severity after regression and examined chromosomal microarray data to see whether children who lost skills shared a different genetic profile from those who did not.

The work drew on the EXPECT cohort at Robert-Debré Hospital, a prospective study of children and adolescents under 17 with autism spectrum disorder. Of 505 children included after exclusions for missing data, 74, or 15%, showed developmental regression, with the average age at loss of skills just over 21 months. The team found that children who later regressed often had a comparatively stronger early developmental profile, including earlier first words and lower prematurity rates, but their parents also reported concern earlier, usually around the same age that regression emerged.

After regression, the differences between the two groups became clearer. Children with regressive autism were more likely to have functional language impairment, lower non-verbal IQ and poorer adaptive communication. They also showed greater difficulty with social interaction and social affect, alongside more severe restricted and repetitive behaviours. The researchers said this points to a broader and more serious functional impact than in non-regressive autism.

Attempts to predict regression from routine clinical data were far less successful. A decision-tree model had very poor sensitivity, and principal component analysis separated the groups statistically but did not identify a clear set of clinical features driving the split. The authors said this leaves the field without reliable early predictors, despite hints that subtle differences may already be present before regression begins.

The genetic findings were also notable. Among 477 children who had chromosomal microarray analysis, the rate of pathogenic or likely pathogenic copy number variants was similar in the regressive and non-regressive groups. But the specific genes affected differed. In the regression group, deletions included NRXN1 and SHANK3, genes long tied to synaptic development and autism, as well as a complete trisomy 21 in one child. In the non-regressive group, the abnormalities were more diverse and included multiple deletions and duplications across different chromosomal regions.

Gene-enrichment analysis suggested that deleted regions in regressive autism were more often linked to oxidative stress, angiogenesis, inflammation and type I interferon signalling, as well as synaptic pathways involving neurexin and neuroligin. That pattern fits with earlier research. A 2019 review of genetic mechanisms in developmental regression said about one-third of children with autism experience regression and found that several autism-linked genes are associated with it, especially genes involved in transcriptional and synaptic regulation. More recently, animal studies have also suggested that inflammation can worsen behavioural problems in Shank3-related autism models, reinforcing the idea that immune and synaptic mechanisms may intersect in some children.

The authors caution that the study does not prove a single biological cause for regression. Chromosomal microarray testing has limits, because it detects large copy number changes rather than smaller sequence variants or epigenetic effects, and the clinical definition of regression still varies across studies. Even so, the findings add weight to the view that regressive autism may deserve to be treated as a more specific clinical and biological entity, and that future work should focus on better early markers and immune-related pathways.