Persistent brain network signatures in psychosis: implications for diagnosis, prognosis, and treatment

Psychotic disorders result in large societal burden given their early onset and chronicity of highly disabling symptoms. For most individuals, the first episode of psychosis is followed by recurring relapses over the lifetime. This chronic course is associated with deteriorating response to treatment and accumulation of disability. Furthermore, during the chronic phase of the illness, individuals with psychotic disorders are at greater risk of dementia, medical comorbidities, and premature mortality. There is clear agreement in the field that comprehensive specialized treatment is critical during the early psychosis phase – typically referring to the 5 years following the first diagnosis – to mitigate disability during the rest of the course of illness, also known as the chronic phase. Similarly, understanding the neurobiological underpinnings of the changes occurring over the course of illness, especially transitioning from the early into the chronic phase, is necessary to develop targeted preventive and treatment strategies. This issue of Neuropsychopharmacology contains work by Foster et al. [1] comparing functional brain networks implicated in psychotic disorders between the early and the chronic phases.

Previous studies of chronic psychosis show functional hypoconnectivity within and between the default mode, frontoparietal, ventral attention, and somatosensory networks, as well as subcortical structures, compared to healthy controls as noteworthy features [2]. Similar changes in functional connectivity have also been reported in first episode psychosis and early psychosis [2]. However, these patterns of functional disruptions have not yet been directly compared across early and chronic stages. Prior analyses have largely relied on univariate, seed-based analyses. In recent years, data-driven, brain-based predictive models, including connectome-based predictive models, have increasingly been used in neuroscience and psychiatry. These models capture multivariate, whole-brain associations between imaging-derived brain measures and behavioral outcomes, advancing our mechanistic understanding of the neural bases of behaviors and illnesses [3].

The current study used connectome-based predictive models to predict and understand the brain networks that underlie symptom severity in individuals with psychosis. For the comparison in this work between early and chronic psychosis, two separate samples were used. The Human Connectome Project for Early Psychosis, a US-based sample of individuals with affective and non-affective psychosis within 5 years of diagnosis, was used as the early psychosis sample. The Strategic Research Program for Brain Sciences, a Japanese sample of individuals with psychosis an average of ~16 years older than the early sample, was the chronic psychosis sample. The authors report a link between individual variations in functional connectivity and the severity of both positive and negative symptoms in both samples. While similar brain networks were implicated with these symptoms in both the early and chronic psychosis samples, the associations were stronger in the chronic psychosis sample, potentially reflecting a greater involvement of those networks with disease progression. Although the functional networks associated with symptom severity were dispersed throughout the brain, the frontoparietal networks, which are involved in executive function and goal-directed behaviors, were the most important.

These results suggest that disruptions in functional network connectivity associated with positive and negative symptoms in early psychosis may become more pronounced over time. However, these findings could also be influenced by other differences between the two samples in factors such as ethnicity, diagnostic makeup, psychiatric and medical history, cumulative antipsychotic exposure, optimization of antipsychotic treatment, use of concomitant medications, socioeconomic status, and/or social isolation [4]. Nevertheless, in the absence of longitudinal data, these data are valuable as a first observation into the changes occurring in the connectome as individuals with psychosis move from the early to the chronic phase of illness.

In the present study, models trained to predict positive symptoms in the early psychosis sample generalized to the chronic psychosis sample. This replication argues for the validity of the results and suggests shared neurobiological bases for psychotic disorders across settings and disease stage. If indeed the greater effect sizes of the predictions in the chronic sample reflect disease progression a number of follow-up questions emerge: Could early interventions targeting these disruptions change the course of illness? Could the intensification of these disruptions during the early psychosis period serve as a prognostic marker? What changes are driven by potentially modifiable (e.g., relapse) or unmodifiable (e.g., aging) factors? Experimental research using interventional approaches (e.g., transcranial magnetic stimulation) to target these disruptions could address the first question, while computational analyses of longitudinal data from participants with early psychosis who do and do not develop chronic psychosis could address the second. The third question would require moving from cross-sectional studies – such as the Human Connectome Project for Early Psychosis – to longitudinal designs, as has already happened for earlier phases of the illness with the Psychosis Risk Outcomes Network (ProNET) and Early Psychosis Intervention Network (EPINET) projects, which study the prodrome and the early phase of illness respectively. Integrating interventional and computational approaches could be used to personalize prevention and treatment strategies over the course of illness based on individual patterns of functional network disruptions. Even when this may be challenging to perform at scale with functional MRI (fMRI), these findings could inform biomarker development with easier to scale tools like functional near-infrared spectroscopy (fNIRS) or electroencephalography (EEG).

Brain-based markers are objective, quantifiable measures that reflect a biological or clinical process. These markers can inform clinical practice and lead to improved diagnostic and prognostic tools for psychosis, allowing for earlier and more targeted interventions. Before the current work can inform clinical practice, these findings must be replicated through prospective, longitudinal studies and validated across diverse clinical samples. Importantly, to establish a brain-based marker for the conversion from early to chronic psychosis, biomarker validity and disease specificity need to be established [5]. Together, these markers will inform the development of novel treatments that target brain activity in these networks to alleviate symptoms and improve long-term outcomes. Nonetheless, this study presents exciting findings highlighting the functional brain networks involved in positive and negative symptoms in early and chronic psychosis, which should be followed upon in future research.

Author contributions

ED and JR drafted and edited this manuscript.

Funding

This work was supported by the following sources: Northwell Health Advancing Women in Science and Medicine (Career Development Award to ED and Educational Achievement Award to ED), Feinstein Institutes for Medical Research (Emerging Scientist Award to ED), Brain and Behavior Research Foundation (Young Investigator Grant to ED), National Institutes of Mental Health (K23MH127300 to JR). Any opinions or recommendations expressed here are those of the authors and do not necessarily reflect the views of the funders.

Competing interests

JR has received research support from Alkermes, advisory board compensation from TEVA, Karuna, Janssen, royalties from UpToDate. ED reports no disclosures.