Endogenous Bold Signal Complexity of Ventromedial Prefrontal Cortex Is Different in Autistic Men Versus Women and Differentially Associated with Compensatory Behavioral Camouflaging

Background: Longstanding theories suggest that there may be an imbalance of excitation versus inhibition (E/I) in the autistic brain. E/I imbalance can result in increased neural noise with long-range dependence and this may be represented in the 1/f signal of non-invasive in-vivo measurements of brain function such as resting state fMRI (rsfMRI). The 1/f signal has been characterised with the Hurst exponent (H), indicating the temporally persistent behaviour (long memory) and stationarity of the signal when 0.5<H<1. However, rsfMRI BOLD signal exhibits non-stationary behaviour that also needs to be quantified. Thus we have utilised a novel approach for estimating H based on the fractionally integrated process (FIP) model to allow H>1, as a non-stationarity index. Using H as a metric of endogenous BOLD signal complexity, we examined how H may differ between males versus females with autism, by expanding on previous work on adult males (Lai et al., 2010). Here, we assess if H exhibits sex-by-diagnosis interaction effects when examined in a 2x2 design.

Objectives: To assess sex-by-diagnosis interactions in H and assess any sex-differential associations with clinical measures of behavioural camouflaging.

Methods: Age-matched male and female adults with ASD (males n=23; females n=25) and a typically developing (TD) group (males n=29; females n=33) were scanned with rsfMRI (3T, TR=1302 ms, 620 volumes) while awake with eyes closed. Data were preprocessed using a standard pipeline in AFNI, along with motion regression, CSF regression and wavelet despiking. Mean time-series from 180 parcels from the HCP parcellation (Glasser et al., 2016) were used to estimate H based on the FIP model. For each region, a linear model was constructed to examine main effects of sex and diagnosis along with the sex-by-diagnosis interaction, after covarying for mean framewise displacement and full-scale IQ. Multiple comparison correction was implemented with FDR q<0.05. A behavioural camouflaging index was computed identical to past work (Lai et al., 2017) and was utilised in correlation analyses with H, stratified by sex within autistic individuals.

Results: A region in ventromedial prefrontal cortex (vMPFC) showed evidence of a sex-by-diagnosis interaction (F=15.134, p=1.76e-4, partial η²=0.12) with a TD>Autism effect in males (d=1.30) but no strong effect in females (d=-0.27). Increasing vMPFC H was significantly associated with increasing behavioural camouflaging in females (r=0.60, p=0.0015), whereas no association was present in males (r=-0.10, p=0.63). The strength of these vMPFC H-camouflaging correlations was different between males versus females with autism (z=2.58, p=0.0098).

Conclusions: These results suggest that an important node of the ‘social brain’, vMPFC, shows lower H in men but not women with autism, compared to same-sex TD controls. Lower H could indicate enhanced neural noise and may impact several key functions of vMPFC involved in social and self-referential cognition, particularly in males with autism. This marker also explains variance in clinically relevant phenomena such as camouflaging, in a sex-differential manner, suggesting that the less neural noise present in vMPFC of females with autism may be related to enhanced ability to camouflage impairments in the social-communicative domain.