Atypical Brain Network Connectivity Proximal to Behavioral Symptom Onset in Toddlers with Autism Spectrum Disorders

Background: Behavioral signs of autism spectrum disorders (ASDs) appear in the first years of life; however, little is known about brain networks organization in ASDs during this critical developmental period, proximal to autism symptom onset. While there is evidence of early anatomical abnormalities in toddlers with ASDs, including increased cortical surface and brain volume, as well as extra-axial cerebrospinal fluid, less is known about functional network organization early in life in ASDs.

Objectives: To examine functional connectivity patterns in toddlers with (or suspected of having) ASD, utilizing multiband functional MRI data acquired during natural sleep. We hypothesized to find atypical maturational trajectories of network integration (strengthening connectivity within networks) and differentiation (decreasing connectivity between networks) in children with ASDs.

Methods: Participants were young children with ASDs (mean age: 28±7 months) and typically developing (TD) toddlers (26±7 months) continuously enrolling in the ongoing longitudinal study of early brain markers in ASDs. Clinical best estimates of ASD diagnoses were established based on DSM-5 criteria, and supported by the Autism Diagnostic Observation Schedule (ADOS-2), with final diagnoses confirmed longitudinally. The Mullen Scales of Early Learning and Vineland Adaptive Behavior Scales were administered to assess visual reception, motor, language, and social development. To date, T1-weighted anatomical MRI and fMRI data were acquired in 22 toddlers with ASDs and 22 TD toddlers during natural sleep. Group Independent Component Analysis (ICA) was applied to fMRI data, and after identifying 10 independent components (ICs) as non-artifact intrinsic functional connectivity networks (iFNs), mean time series were extracted from thresholded iFN masks and entered into a 10x10 pairwise correlation matrix, for each participant. Group differences in iFN connectivity were examined with two-sample t-tests applied to the correlation matrices.

Results: As expected, participants with ASDs showed delayed development across multiple behavioral domains including gross and fine motor skills, visual reception, receptive and expressive language, and socialization. Group comparisons of iFNs revealed significantly weaker connectivity between medial and lateral visual iFNs (p=0.03, FDR corrected), as well as significantly stronger connectivity between medial visual iFN and sensorimotor network (p=0.01, FDR corrected) in toddlers with ASDs. Connectivity between the medial and lateral visual RFNs was negatively correlated with age in ASD (r=-0.66) but not in TD (r=0.03) toddlers, with significant age by diagnosis interaction. In toddlers with ASDs, greater connectivity between visual and motor networks was correlated with greater symptom severity measured with ADOS (r=0.58, p=0.02) and with worse performance on the Mullen Early Learning Scales (r=-0.50, p=0.04).

Conclusions: These findings point to atypical functional connectivity implicating visual network in 1.5-3.5 years old children with ASDs, including weaker connectivity within visual networks, and stronger connectivity between visual and motor regions. The age-related decline in the connectivity between visual and other primary sensorimotor networks observed in ASDs between age 1.5 and 3.5 years, but not detected in TD children, suggests a maturational delay in network segregation in ASDs. This interpretation was supported by the data indicating that greater network segregation in toddlers with ASDs was associated with decreased symptoms and greater developmental scores.

Background: Behavioral signs of autism spectrum disorders (ASDs) appear in the first years of life; however, little is known about brain networks organization in ASDs during this critical developmental period, proximal to autism symptom onset. While there is evidence of early anatomical abnormalities in toddlers with ASDs, including increased cortical surface and brain volume, as well as extra-axial cerebrospinal fluid, less is known about functional network organization early in life in ASDs.

Objectives: To examine functional connectivity patterns in toddlers with (or suspected of having) ASD, utilizing multiband functional MRI data acquired during natural sleep. We hypothesized to find atypical maturational trajectories of network integration (strengthening connectivity within networks) and differentiation (decreasing connectivity between networks) in children with ASDs.

Methods: Participants were young children with ASDs (mean age: 28±7 months) and typically developing (TD) toddlers (26±7 months) continuously enrolling in the ongoing longitudinal study of early brain markers in ASDs. Clinical best estimates of ASD diagnoses were established based on DSM-5 criteria, and supported by the Autism Diagnostic Observation Schedule (ADOS-2), with final diagnoses confirmed longitudinally. The Mullen Scales of Early Learning and Vineland Adaptive Behavior Scales were administered to assess visual reception, motor, language, and social development. To date, T1-weighted anatomical MRI and fMRI data were acquired in 22 toddlers with ASDs and 22 TD toddlers during natural sleep. Group Independent Component Analysis (ICA) was applied to fMRI data, and after identifying 10 independent components as non-artifact intrinsic functional connectivity networks (iFNs), mean time series were extracted from thresholded iFN masks and entered into a 10x10 pairwise correlation matrix, for each participant. Group differences in iFN connectivity were examined with two-sample t-tests applied to the correlation matrices.

Results: As expected, participants with ASDs showed delayed development across multiple behavioral domains including gross and fine motor skills, visual reception, receptive and expressive language, and socialization. Group comparisons of iFNs revealed significantly weaker connectivity between medial and lateral visual iFNs (p=0.03, FDR corrected), as well as significantly stronger connectivity between medial visual iFN and sensorimotor network (p=0.01, FDR corrected) in toddlers with ASDs. Connectivity between the medial and lateral visual RFNs was negatively correlated with age in ASD (r=-0.66) but not in TD (r=0.03) toddlers, with significant age by diagnosis interaction. In toddlers with ASDs, greater connectivity between visual and motor networks was correlated with greater symptom severity measured with ADOS (r=0.58, p=0.02) and with worse performance on the Mullen Early Learning Scales (r=-0.50, p=0.04).

Conclusions: These findings point to atypical functional connectivity implicating visual network in 1.5-3.5 years old children with ASDs, including weaker connectivity within visual networks, and stronger connectivity between visual and motor regions. The age-related decline in the connectivity between visual and other primary sensorimotor networks observed in ASDs between age 1.5 and 3.5 years, but not detected in TD children, suggests a maturational delay in network segregation in ASDs. This interpretation was supported by the data indicating that greater network segregation in toddlers with ASDs was associated with decreased symptoms and greater developmental scores.