Risk and resilience brain networks in treatment-resistant schizophrenia

Abstract

Background

Genes, molecules and neural circuits that are associated with, or confer risk to developing schizophrenia have been studied and mapped. It is hypothesized that certain neural systems may counterbalance familial risk of schizophrenia, and thus confer resilience to developing the disorder. This study sought to identify resting-state functional brain connectivity (rs-FC) representing putative risk or resilience endophenotypes in schizophrenia.

Methods

Resting-state functional magnetic resonance imaging (rs-fMRI) was performed in 42 individuals with treatment resistant schizophrenia (TRS), 16 unaffected first-degree family members (UFM) and 42 healthy controls. Whole-brain rs-FC networks were mapped for each individual and analysed graph theoretically to identify network markers associated with schizophrenia risk or resilience.

Results

The ~ 900 functional connections showing between-group differences were operationalized as conferring: i) resilience, ii) risk, or iii) precipitating risk and/or illness effects. Approximately 95% of connections belonged to the latter two categories, with substantially fewer connections associated with resilience. Schizophrenia risk primarily involved reduced frontal and occipital rs-FC, with patients showing additional reduced frontal and temporal rs-FC. Functional brain networks were characterized by greater local efficiency in UFM, compared to TRS and controls.

Conclusions

TRS and UFM share frontal and occipital rs-FC deficits, representing a ‘risk’ endophenotype. Additional reductions in frontal and temporal rs-FC appear to be associated with risk that precipitates psychosis in vulnerable individuals, or may be due to other illness-related effects, such as medication. Functional brain networks are more topologically resilient in UFM compared to TRS, which may protect UFM from psychosis onset despite familial liability.

Introduction

Schizophrenia has a strong genetic component, with the most prominent risk factor for developing the disorder being family history (Gottesman and Gould, 2003, Kendler and Neale, 2010). Studying unaffected relatives of individuals with schizophrenia can therefore offer insight into the heritable pathophysiology of the disorder, independent of factors that often confound studies in patients, such as illness progression and chronic antipsychotic use (Braff et al., 2007). A number of structural brain alterations are shared between schizophrenia patients and their unaffected family members (UFM) representing candidate endophenotypes (Moran et al., 2013, Turetsky et al., 2007), such as cortical thinning (Goghari et al., 2007, Gogtay et al., 2007; Yang et al., 2010), reduced morphological covariance (Zalesky et al., 2015), whole brain (McIntosh et al., 2011, Thermenos et al., 2013) and subcortical volume reductions (Peper et al., 2007, Thermenos et al., 2013). Less understood, however, is the nature of structural and functional brain connectivity abnormalities in UFM. Decreased resting-state functional connectivity (rs-FC) is commonly reported in schizophrenia patients, although increased rs-FC is also described (for review, see Fitzsimmons et al., 2013). Similarly, studies in UFM have generated mixed results, with some findings showing increased rs-FC relative to controls (Jang et al., 2011, Jukuri et al., 2013, van Buuren et al., 2012, Whitfield-Gabrieli et al., 2009), while others report reduced rs-FC (A. Fornito et al., 2013, Jang et al., 2011, Jukuri et al., 2015, Khadka et al., 2013; Liu et al., 2012, Meda et al., 2012, Sole-Padulles et al., 2016). On the whole, rs-FC is predominantly reduced in schizophrenia patients (Fornito et al., 2012), and the functional brain networks most affected often show milder rs-FC reductions in UFM (Wang et al., 2015). Aberrant rs-FC networks shared between patients and UFM may therefore represent a marker of genetic vulnerability to schizophrenia, rather than solely the result of illness duration, medication and/or other secondary environmental factors (Repovs et al., 2011).

Alternatively, rs-FC alterations that are unique to UFM and absent or moderated in affected relatives and the general population might be hypothesized to represent putative markers of resilience to schizophrenia, and counterbalance familial liability. Resilience biomarkers have not been extensively studied in schizophrenia, with resilience in psychiatry traditionally broached in terms of psychological response to stress and trauma (Feder et al., 2009, Russo et al., 2012). Recent evidence suggests that functional brain networks in UFM show increased resilience to pathological disruptions, compared to schizophrenia patients and controls (Lo et al., 2015). Similarly, UFM show resilience in that they recover from developmental delays in structural connectivity (Chakravarty et al., 2015, Zalesky et al., 2015). Resilience endophenotypes inferred from rs-FC have also been reported in depression (Peterson et al., 2014). These previous studies motivate investigation of functional brain networks associated with resilience in schizophrenia.

Here, we perform resting-state functional magnetic resonance imaging (fMRI) in individuals with treatment-resistant schizophrenia (TRS), UFM and healthy controls, with the aim of identifying functional brain networks associated with schizophrenia risk or resilience. We operationalize resilience as functional connections or functional network properties that are uniquely present (or absent) in UFM individuals, whereas network properties that are shared between TRS and UFM (but not evident in the general population) are considered risk markers. We consider TRS patients in this study to ensure a homogeneous clinical phenotype (Jablensky, 2006), and thereby maximize the reproducibility of our findings. We and others have found that TRS is associated with widespread abnormal rs-FC (Ganella et al., 2017, Vercammen et al., 2010, White et al., 2016, Wolf et al., 2011) and we hypothesize that investigating UFM of TRS patients will provide insight into functional networks associated with schizophrenia risk and resilience. Specifically, we hypothesize both the TRS and UFM groups to show reduced rs-FC and network efficiency relative to controls, albeit to a lesser extent in UFM.