Functional Connectivity Alterations of the Prefrontal Thalamic Pathway in Autism Spectrum Disorder Are Age and Sex Dependent

Background:

The thalamus maintains connections throughout the brain and its dysfunction has been associated with autism. However, DTI studies investigating structural functional connectivity are inconsistent and little is known about subregion-level thalamic functional connectivity (FC) in ASD. Therefore, we investigated subregion thalamo-cortical FC strength in ASD using a discovery sample and four replication samples of the ABIDE I dataset. The replication samples varied from our discovery sample with respect to sample size, scanning site and imaging parameters, sex-ratio, and the age range studied.

Objectives:

To delineate robust thalamic subregion abnormalities in ASD, we aimed to reproduce our findings from an adolescent discovery sample across four distinct replication samples.

Methods:

In the discovery sample (ASD=19 male/1 female, 12-22y; CTR=22male/3female, 13-22y), using an age and gender corrected correlation analysis, we calculated the full-correlations between four thalamic subregions (prefrontal (PF-T), temporal (TE-T), parietal-occipital (OP-T) and sensory, primary motor and pre-motor (SM-T)) and their associated cortical lobes separately in resting-state fMRI scan data. We identified increased FC strength along the right prefrontal thalamic nucleus (PF-T) and the right ventromedial prefrontal cortex (vmPFC). In the planned analysis in the four replication samples (ASD_Leuven29=14 male, 18-32y; CTR_Leuven29=15 male, 18-29y; ASD_Leuven35=12 male/3 female, 12-16y; CTR_Leuven35=15 male/5 female, 12-16y; ASD_USM101=58 male, 11-50y;CTR_USM101=43male, 8-39y; ASD_NYU184=68 male/11 female,7-39y; CTR_NYU184=79 male/26 female, 6-31y) we extracted the mean time series between the right PF-T and vmPFC (PF-T_vmPFC FC strength) and its r-to-Z transformed correlations were calculated. Using lmrob a robust linear modelling method as implemented in the statistical R software package ‘robustbase’, we predicted the effect of age, sex and symptom severity (SRS) or diagnostic group status (Group) on PF-T_vmPFC FC strength.

Results:

The PF-T_vmPFC FC strength effect was dependent on site when investigating the SRS score (χ²(df=1)=8.17, df=3, p=0.043) across all sites, or absent when investigating diagnostic group across all sites. However, we found sex dependent effects in both of the samples that contained females (Leuven35=(χ²(df=1)=28.10, df=1, p<0.0001; NYU184=(χ²(df=1)=4.33, df=3, p=0.037)) when investigating the SRS score, but only in one sample using diagnostic status as predictor (Leuven35=(χ²(df=1)= 12.5, p<0.0001)). In one sample, we found interaction effects with diagnostic group and age (Leuven29= χ²(df=1)=6.06, p<0.015)). In 2 of the 4 samples, there was an effect of age ((Leuven29= χ²(df=1)=5.96, p<0.015; USM101= χ²(df=1)=4.17, p<0.05). In the sample that was the most similar to the discovery sample, Leuven35, we found that sex, age and diagnostic status as well as sex, age and SRS symptom scores were robust predictors of right PF-T_vmPFC FC strength (all p-values <0.001).

Conclusions:

Increased thalamic FC strength in ASD was identified in prefrontal nucleus within our discovery sample. PF-T_vmPFC FC strength was not predicted by SRS scores or diagnostic status across all sites per se, while there were age and sex interaction effects with diagnostic group. These findings suggest that there are age and sex specific developmental thalamic-cortical pathways in the etiology of ASD.