Cluster Analysis of Responsivity to Citalopram in Autism Spectrum Conditions: Preliminary Evidence of Neurobiological Sub-Groups.
Objectives: Therefore, here we tested the hypothesis that there are E-I responsivity differences to selective serotonin reuptake inhibition (SSRI) with citalopram within a group of individuals with ASC and between ASC and neurotypical controls.
Methods: This study used MEGAPRESS proton magnetic resonance spectroscopy ([1H]-MRS) to measure concentrations of GLx (glutamate + glutamine) and GABA in unmedicated adult men with ASD (n=19) and age- and IQ- matched controls (n=20). Individuals were scanned twice, once after oral administration of 20mg of citalopram and once under placebo in a randomised double-blind procedure. Scans were at least 8 days apart to ensure full washout of the drug. Spectra was acquired in the dorsomedial pre-frontal cortex (DMPFC). Responsivity was defined as percentage change (placebo to citalopram) for each metabolite. Two-step cluster analysis was performed in order to identify subgroups based on response.
Results: There was a significant group by drug interaction on Glx levels (p<0.05). Overall Citalopram decreased Glx in controls but had no significant impact on Glx in ASD. However, this result masked marked individual variation in response to citalopram in both groups. Cluster analysis revealed a three-cluster solution with ‘good’ fit (average silhouette = 0.7). Clusters could be characterised as 1) ‘responders with decreased Glx after citalopram’ (ASD=4, Control=6 average response = -18.30%), ‘non-responders’ (ASD=10, Control= 13, average response = -0.71%) and ‘responders with increased Glx after citalopram’ (ASD=6, Control=0, average response = +21.42%). There were no significant interactions or main effects of GABA.
Conclusions: These findings suggest that Glx response to an SSRI is significantly different in ASD compared to controls. The considerable individual variation in pharmacological response which may inform efforts to stratify individuals with ASC into more biological homogeneous subgroups. This may encourage future work to examine whether these responsivity differences are predicative of treatment response.