On the Feasibility of Rt-fMRI Neurofeedback Training in ASD: A Clinical Trial Study

Background: The posterior superior temporal sulcus (pSTS) has been previously identified as part of the face processing core network. It also plays a key role on basic aspects of social information processing, with some imaging studies reporting abnormal function and structure in Autism Spectrum Disorder (ASD).

Advances in functional Magnetic Resonance Imaging (fMRI) allow to feedback quasi real-time information about the brain function to the participant (rt-fMRI-neurofeedback). In this sense, the participant learns to volitionally self-modulate the activation of a specific brain region. It has been demonstrated that neurofeedback training alters neural activity and preliminary studies suggest that the approach may represent a viable neurorehabilitation tool.

Objectives: We investigated the feasibility of a neurofeedback-based intervention in ASD, targeting a brain region involved in social cognition - the pSTS.

We aim to a) evaluate patients’ acceptability and recruitment over multiple sessions, b) assess preliminary efficacy measures derived from neuropsychological evaluation and imaging data.

Methods: 15 participants (mean age 19y11m) with high-functioning ASD (mean full-scale IQ 103) were invited to enroll in a single-arm feasibility clinical trial. The intervention structure consisted of an initial eligibility screening, pre-intervention (first week of study, baseline measures), 5 sessions rt-fMRI-NF intervention process (4 weekly sessions, fifth one month later), post-intervention, and follow-up (at 6 months).

Each neurofeedback session lasted for approximately 1 hour. During this time, the participants were asked to imagine different facial expressions (as a strategy for volitional control of pSTS activity), while looking at a representation of their own brain activity. The primary outcome measure is the FEEST - Hexagon test - and the secondary outcome measures are the Autism Treatment Evaluation Checklist (ATEC) and Vineland Adaptive Behaviour Scale (VABS). Mood and depression state were also assessed. Regarding imaging data, we assessed a general linear model in our region-of-interest (ROI-GLM) throughout the intervention.

Results: 10 out of 15 patients improved in global ability to recognize expressions according to FEEST. The results present a 13% average decrease in total ATEC (Z=-2.076, p=0.038) rated autism symptoms (23% in Sensory/Cognitive Awareness, Z=-2.446, p=0.014; 21% in Health/Physical/Behaviour, Z=-2.176, p=0.030); 6% average improvement in Adapted Behaviour Composite (Z=-2.768, p=0.006) and in all VABS subareas: 6% in Communication (Z=-2.366, p=0.018), 4% in Daily Living Skills (Z=-2.324, p=0.020), 5% in Socialization (Z=-2.491, p=0.013); 9% average decrease in Total Mood Disturbance (Z=-2.103, p=0.035), 44% in Tension (Z=-2.313, p=0.021); 82% in Depression (Z=-2.139, p=0.032), 64% in Anger (Z=-2.286, p=0.022) and 43% average decrease in mood disturbance/depression, Z=-2.421, p=0.015.

The ROI-GLM analysis suggests that pSTS modulation is possible (42.67% of the runs were successful according to two-tailed t-tests, P-val<0.05). We further explored the data, with an individual assessment of the event-related averages curves, and found evidence for heterogeneity (e.g. participants modulated either positively or negatively to distinct regulation conditions).

Conclusions: These findings support the hypothesis that neurofeedback may represent a viable neurorehabilitation tool. Tailored strategy optimization may represent an important aspect in future studies.

Results emphasize the need to proceed to a two-arm randomized clinical trial to validate the reported preliminary efficacy findings.