Excitatory/Inhibitory Balance in the Basal Nuclei Circuit in Treatment-Resistant Schizophrenia: A Computational Modelling Approach Using The Virtual Brain

Abstract Background: Schizophrenia (SZ) is a severe psychiatric disorder with a lifetime prevalence of approximately 1%. Nearly one third of the patients with SZ show little or no improvement in positive symptoms, an excess and/or distortion of typical mental functioning, such as hallucinations and delusions, with first-line antipsychotics. This condition is termed treatment resistant schizophrenia (TRS). Compared with treatment responsive patients (non-TRS), those with TRS have lower quality of life, poorer cognitive and social functioning and higher risk of substance abuse. The clinical heterogeneity motivates the central question of this thesis: whether TRS reflects a distinct pattern of circuit-level dysfunction, rather than simply a more severe version of SZ. Objectives: This thesis had two aims. The first aim was to establish whether the resulting circuit-level parameters, global coupling (G), local inhibitory coupling (Ji), local excitatory NMDA mediated coupling (JNMDA), and recurrent excitation (w+), could distinguish schizophrenia patients from healthy controls (HC), and TRS from non-treatment resistant (non-TRS) patients. The second aim was to test whether these same parameters, together with treatment resistance status, could predict symptom severity, measured using Positive and Negative Syndrome Scale (PANSS) scores, within the SZ group. These parameters were obtained using The Virtual Brain (TVB) to model a basal nuclei circuit and characterise its excitatory/inhibitory (E/I) balance, the relative strengths of excitatory versus inhibitory signalling. The circuit was expanded beyond standard basal nuclei parcellations to include dentate nuclei, cerebellar structures not typically included in basal nuclei parcellations. Methods: The final sample comprised 76 participants. Thirty-three were patients with SZ and 43 were healthy controls. Subject-specific models of the basal nuclei circuit were built for each participant and its brain dynamics were simulated using the Wong-Wang neural mass model in TVB. Results: None of the four basal nuclei circuit parameters, nor of the two fit indices (PCC, KS), or the overall cost function, differed significantly between SZ patients and HCs. The same finding was obtained when comparing TRS, non-TRS, and HCs. When we tested the ability of these parameters to predict symptom severity using multiple regression, treatment resistance status emerged as a consistent predictor of PANSS total, negative and disorganisation scores. Moreover, Ji significantly predicted disorganisation symptom severity at an uncorrected threshold. Conclusion: Taken together, these findings do not establish a new biomarker of treatment resistance in SZ. However, they do provide a preliminary basis for future work relating basal nuclei circuit dynamics to the clinical heterogeneity of SZ. Independent of these findings, the dentate-inclusive parcellation developed in this work constitutes a methodological contribution in its own right. Keywords: Schizophrenia, Treatment Resistant Schizophrenia, The Virtual Brain, Basal Nuclei, Dentate Nucleus