Hyperactivation of the Posterior Medial Network Supports Preserved Episodic Memory in Adolescents with Autism Spectrum Disorder

Background:

There is ongoing debate about whether memory is spared or impaired in autism spectrum disorder (ASD). Most research suggest that the deficits found in ASD are primarily related to episodic memory, and include impairments in relational encoding and recollection. Previously, our group published a behavioral study using the well-validated Relational and Item-Specific Encoding (RiSE) paradigm (Solomon, McCauley, Iosif, Carter, & Ragland, 2016). See Figure 1 (A). This task is premised on the notion that item-specific encoding and familiarity are sub-served by an anterior-temporal (AT) network anchored in the perirhinal cortex of the medial temporal lobe (MTL), whereas relational encoding and the process of recollection rely on a posterior-medial (PM) network anchored by the hippocampus and parahippocampal cortex. Contrary to our predictions, during a recollection test, ASD versus TYP showed lower discriminability for items encoded in the item-specific versus the relational condition; used familiarity less than recollection for items encoded relationally; and showed greater improvements in performance with age. To develop a mechanistic explanation for these findings, we conducted a well-powered functional magnetic resonance imaging (fMRI) study of the RiSE task in adolescents and young adults with ASD.

Objectives:

Study goals were to investigate whether: 1) relational encoding is impaired in ASD, 2) whether impairment is associated with aberrant recruitment of the PM versus the AT network, and whether 3) patterns of neural recruitment change from adolescence to young adulthood.

Methods:

One-hundred and seven individuals with ASD (N=47) or typical development (TYP, N=60) from the completed first wave of an NIMH-funded 5 year cohort-sequential study on development of cognitive control in adolescents and young adults ages 12-22 years performed the RiSE task while undergoing fMRI. Accuracy (d’) during item recognition (IR-d’) and associative recognition (AR-d’) indexed performance, and robust inferential analyses were deployed. Threshold-free cluster enhancement (TFCE) was used to isolate brain activation within the PM and AT networks, and functional connectivity with false-discovery rate (FDR) correction was used to elucidate interactions between regions.

Results:

IR-d’ and AR-d’ did not differ between ASD and TYP (p≥0.607; BF₀₁≥3.80). See Figure 1 (B, C). Despite similar performance, the ASD group demonstrated PM hyperactivation compared to TYP (p_TFCE<0.025), which was associated with IR-d’ in ASD (rho=0.36, p=0.006). In contrast, PM activity during relational encoding was associated with AR-d’ in TYP (rho=0.64, p<0.001). Lastly, PM network connectivity was greater in TYP relative to ASD (p_FDR<0.05), and increased regional activation of PM clusters was negatively associated with PM network connectivity in ASD (rho=-0.33, p=0.014). See Figure 2. The aforementioned findings were not associated with age in either group.

Conclusions:

The present results suggest that, through adolescence and early adulthood, increased local activation within PM regions may be a compensatory mechanism for decreased PM network connectivity during relational encoding in ASD, ultimately supporting broadly similar episodic memory performance in both groups. Accordingly, interventions designed to facilitate PM network connectivity may represent a viable personalized medicine approach for cognitive treatment in adolescents and young adults with ASD who exhibit relational encoding impairments.