Relationship Between MR Spectroscopy-Detected Glutamatergic Neurometabolites and Changes in Social Behaviors in a Pilot Open-Label Trial of Memantine for Adults With Autism Spectrum Disorder

BACKGROUND: The neurobiology underlying ASD is largely unknown but altered neural excitability/inhibitory ratios have been reported. Memantine is an N-methyl-D-aspartate (NMDA) glutamatergic antagonist studied for the treatment of core ASD symptoms, with mixed results. We examined whether glutamatergic levels were associated with and predicted response to memantine in an exploratory pilot study. METHODS: Ten adult participants with ASD underwent proton magnetic resonance spectroscopy ((1)H-MRS) imaging at baseline and behavioral assessments before and after 12-weeks of open-label memantine. Post-treatment scores on Clinical Global Impressions–Improvement (CGI-I) for social interaction were the primary outcome measure, and scores on the Social Responsiveness Scale (SRS) were included as a secondary outcome. LCModel was used to quantify the concentrations of Point RESolved Spectroscopy-detected glutamate+glutamine (Glx) (and other neurometabolites, i.e., N-acetylaspartate, NAA; creatine+phosphocreatine, Cr+PCr, and myo-inositol, Ins), within the left dorsolateral prefrontal cortex (LDLPFC) and right (R) posterolateral cerebellum. SPM was used to perform brain tissue segmentation within the spectroscopic voxels. CGI-I scores post-treatment were used to classify the participants into two groups, responders (scores 1–3; n = 5) and non-responders (scores 4–7, or withdrew due to increase behaviors; n = 5). Independent samples t-tests, partial correlations and linear hierarchical regression models (SPSS) were used to determine between-group differences in neurometabolite concentrations and associations between neurometabolites and behavioral scores. RESULTS: Responders and non-responders did not significantly differ in Glx levels in any region of interest, but differed in NAA levels in LDLPFC (higher in responders vs. non-responders). Although changes in CGI-I social scores were not correlated with Glx in any region of interest, the linear hierarchical regression did reveal that Glx and Ins levels in LDLPFC were predictors of post-treatment CGI-I social scores. Changes in SRS scores were correlated with baseline Cr+PCr levels in the LDLPFC. DISCUSSION: Our pilot data suggest that baseline Glx, a marker of glutamatergic neurotransmission, did not directly predict response to memantine for social outcomes in adults with ASD. However, interactions between Glx and the neurometabolite associated with glial integrity (Ins) may help predict treatment response. Further, those with highest baseline NAA, a putative neuronal marker, and Cr+pCr, a brain energy metabolism marker, were the best responders. These preliminary results may explain some of the mixed results reported in previous memantine trials in ASD. Future studies will need to examine these results in a larger sample.