O11.1. ASSOCIATION BETWEEN GLUTAMATE AND ACCELERATED BRAIN-AGEING IN TREATMENT-RESISTANT SCHIZOPHRENIA

Abstract

Background: Previous studies have shown that glutamatergic dysfunction appears to be a stable neurobiological characteristic of antipsychotic treatment-resistance. At the same time, treatment-resistant patients show more pronounced neurostructural deficits when compared to treatmentresponsive patients. However, this association between glutamate and neurostructural integrity has not been investigate before in patients with treatment-resistant schizophrenia. Method(s): 42 participants with a diagnosis of schizophrenia and 20 healthy controls matched for age, sex, and socioeconomic background consented to participate in this study. We acquired T1-weighted magnetization prepared rapid acquisition gradient echo sequence and proton magnetic resonance spectroscopy at 3 Tesla from bilateral anterior cingulate cortex. The patient group was classified into 21 antipsychotic treatment-resistant patients and 21 antipsychotic treatment-responsive patients. To calculate brain-age we used a novel method termed brain-predicted age difference (brain-PAD; Cole et al., 2016). This method uses a machine-learning algorithm to calculate brain-age calculated from structural images. The model was trained on structural images from a large healthy reference sample (N=2001), which were used to predict chronological age in a Gaussian Process regression. The model estimates were used to predict brain-PAD in the participants of our study independently, using their T1 structural scans. Brain-PAD was calculated as brain-predicted age minus chronological age. Brain-PAD scores above zero indicate accelerated brain ageing. Using the T1-weighted MRI scans we also calculated cortical thickness using the standard reconall pipeline available in Freesurfer 6.0 software. Cortical thickness was calculated over 34 regions of interest in each hemisphere based on the standard Desikan-Killiany atlas. 1H-MRS data were analyzed using a standard basis function within LC-Model. Our primary measure was glutamate to creatine ratio (Glu/Cre). Result(s): The main finding is that there was a significant positive correlation between anterior cingulate cortex Glu/Cre and brain-PAD only for the treatment-resistant patients, R=0.52, P=0.014, meaning that within this group higher Glu/Cre values were associated with accelerated brain ageing. There was no significant correlation for the treatment-responsive group, R=-0.23, P=0.33 or the healthy participants, R=0.09, P=0.78. Fisher's r-toz transformations to test pair-wise group differences showed that Glu/Cr and brain-PAD correlation was significantly stronger in treatment-resistant patients when compared to treatment-responsive, Z=2.36, P=0.018 reflecting a large effect size of 0.81. The difference between treatment-resistant and healthy participants was at trend-level, Z=1.36, P=0.08, with a medium effect size of 0.49. There was no difference between treatment-responsive and healthy participants, Z=-0.92, P=0.36. A regression model including cortical thickness showed that for treatment-resistant patients, increase in strength of the relationship between Glu/Cre and brain-PAD was associated with cortical thickness reductions in right medial temporal lobe, bilateral inferior parietal lobe, right superior parietal lobe and precuneus Discussion: We show for the first time that glutamate levels were associated with accelerated brain ageing in treatment-resistant patients. This is consistent with accounts of glutamate-mediated excitotocity. Our data suggest that glutamatergic dysfunction in patients with poor antipsychotic response can particularly affect neurostructural integrity in cortical regions associated with the pathoetiology of the disorder.