Linked dimensions of psychopathology and connectivity in functional brain networks.
Xia CH, Ma Z, Ciric R, Gu S, Betzel RF, Kaczkurkin AN, Calkins ME, Cook PA, García de la Garza A, Vandekar SN, Cui Z, Moore TM, Roalf DR, Ruparel K, Wolf DH, Davatzikos C, Gur RC, Gur RE, Shinohara RT, Bassett DS, Satterthwaite TD. · 2018
Read Linked dimensions of psychopathology and connectivity in functional brain networks. by Xia CH, Ma Z, Ciric R, Gu S, Betzel RF, Kaczkurkin AN, Calkins ME, Cook PA, García de la Garza A, Vandekar SN, Cui Z, Moore TM, Roalf DR, Ruparel K, Wolf DH, Davatzikos C, Gur RC, Gur RE, Shinohara RT, Bassett DS, Satterthwaite TD. (2018) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc6070480.txt.
Introduction
Xia and co-authors link dimensions of mental illness to brain connectivity in youth. Their study appeared in Nature Communications. Diagnoses map poorly onto brain biology. The team instead let symptoms and scans define dimensions. They studied 663 young people with a sparse correlation method.
Four dimensions emerged from the data. They covered mood, psychosis, fear, and externalizing behavior. Each tied to a distinct connectivity pattern. All four shared weaker separation between two major networks. Mood and psychosis patterns grew clearer with age. Sex differences appeared for mood and fear. Findings mostly repeated in a second sample of 336. Readers get brain-based dimensions that cross diagnosis lines.
Catalog record
| CREATOR | Xia CH, Ma Z, Ciric R, Gu S, Betzel RF, Kaczkurkin AN, Calkins ME, Cook PA, García de la Garza A, Vandekar SN, Cui Z, Moore TM, Roalf DR, Ruparel K, Wolf DH, Davatzikos C, Gur RC, Gur RE, Shinohara RT, Bassett DS, Satterthwaite TD. |
|---|---|
| YEAR | 2018 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 79 KB |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Linked dimensions of psychopathology and connectivity in functional brain networks By Cedric Huchuan Xia, Zongming Ma, Rastko Ciric, Shi Gu, Richard F. Betzel, Antonia N. Kaczkurkin, Monica E. Calkins, Philip A. Cook, Angel García de la Garza, Simon N. Vandekar, Zaixu Cui, Tyler M. Moore Published in: Nature Communications DOI: 10.1038/s41467-018-05317-y ABSTRACT Neurobiological abnormalities associated with psychiatric disorders do not map well to existing diagnostic categories. High co-morbidity suggests dimensional circuit-level abnormalities that cross diagnoses. Here we seek to identify brain-based dimensions of psychopathology using sparse canonical correlation analysis in a sample of 663 youths. This analysis reveals correlated patterns of functional connectivity and psychiatric symptoms. We find that four dimensions of psychopathology – mood, psychosis, fear, and externalizing behavior – are associated (r = 0.68–0.71) with distinct patterns of connectivity. Loss of network segregation between the default mode network and executive networks emerges as a common feature across all dimensions. Connectivity linked to mood and psychosis becomes more prominent with development, and sex differences are present for connectivity related to mood and fear. Critically, findings largely replicate in an independent dataset (n = 336). These results delineate connectivity-guided dimensions of psychopathology that cross clinical diagnostic categories, which could serve as a foundation for developing network-based biomarkers in psychiatry. INTRODUCTION Psychiatry relies on signs and symptoms for clinical decision making, without the use of established biomarkers to aid in diagnosis, prognosis, and treatment selection. It is increasingly recognized that existing clinical diagnostic categories could hinder the search for biomarkers in psychiatry1, as they are not clearly associated with distinct neurobiological abnormalities2. The high co-morbidity among psychiatric disorders exacerbates this problem3. Furthermore, studies have demonstrated common structural, functional, and genetic abnormalities across psychiatric syndromes, potentially explaining such co-morbidity4–6. This body of evidence underscores the lack of direct mapping between clinical diagnostic categories and the underlying pathophysiology. This context has motivated the development of the National Institute of Mental Health’s Research Domain Criteria, which seek to construct a biologically-grounded framework for psychiatric diseases7. In such a model, the symptoms of individual patients are conceptualized as the result of mixed dimensional abnormalities of specific brain circuits. While such a model system is theoretically attractive, it has been challenging to implement in practice due to both the multiplicity of clinical symptoms and the many brain systems implicated in psychiatric disorders. Network neuroscience is a powerful approach for examining brain systems implicated in psychopathology8,9. One network property commonly evaluated is its community structure, or modular architecture. A network module (also called a sub-network or a community) is a group of densely interconnected nodes, which may form the basis for specialized sub-units of information processing. Converging results across data sets, methods, and laboratories provide substantial agreement on large-scale functional brain modules such as the somatomotor, visual, default mode, and fronto-parietal control networks10–12. Furthermore, multiple studies documented abnormalities within this modular topology in psychiatric disorders13,14. Specifically, evidence suggests that many psychiatric disorders are associated with abnormalities in network modules subserving higher-order cognitive processes, including the default mode and fronto-parietal control networks15,16. In addition to such module-specific deficits, studies in mood disorders17,18, psychosis14,19, and other disorders20,21 have reported abnormal interactions between modules that are typically segregated from each other at rest. This is of particular interest as modular segregation of both functional22,23 and structural24 brain networks is refined during adolescence, a critical period when many psychiatric disorders emerge. Such findings have led many disorders to be considered “neurodevelopmental connectopathies.”25 Describing the developmental substrates of psychiatric disorders is a necessary step towards early identification of at-risk youth, and might ultimately allow for interventions that “bend the curve” of maturation to achieve improved functional outcomes26. Despite the increasing interest in describing how abnormalities of brain network development lead to the emergence of psychiatric disorders, existing studies have been limited in several respects. First, most have adopted a categorical case-control approach, or only examined a single dimension of psychopathology15, and are therefore unable to capture heterogeneity across diagnoses. Second, dimensional psychopathology derived from factor analyses, including our prior work27–30, were solely driven by covariance in the clinical symptomatology, rather than being guided by both brain and behavior features. Third, especially in contrast to adult studies, existing work in youth has often used relatively small samples (e.g., dozens of participants). While multivariate techniques allow the examination of both multiple brain systems and clinical dimensions simultaneously, such techniques usually require large samples31. In the current study, we seek to delineate functional network abnormalities associated with a broad array of psychopathology in youth. We have capitalized on a large sample of youth from the Philadelphia Neurodevelopmental Cohort (PNC)32 by applying a recently-developed machine learning technique called sparse canonical correlation analysis (sCCA)33. As a multivariate method, sCCA is capable of discovering complex linear relationships between two high-dimensional datasets34,35. It should be noted that the approach of the current study is distinct from prior work discovering biotypes within categories of psychopathology, based purely on imaging features themselves (e.g., functional connectivity36 and gray matter density37). In contrast, we seek to link a broad range of symptoms that are present across categories to individual differences in functional brain networks. Such an approach has been successfully applied in prior work on neurodegenerative diseases34 as well as normal brain-behavior relationships35. Here, we use sCCA to delineate linked dimensions of psychopathology and functional connectivity. As described below, we uncover dimensions of connectivity that are highly correlated with specific, interpretable dimensions of psychopathology. We find that each psychopathological dimension is associated with a distinct pattern of abnormal connectivity, and that all dimensions are characterized by decreased segregation of default mode and executive networks (fronto-parietal and salience). These network features linked to each dimension of psychopathology show expected developmental changes and sex differences. Finally, our results are largely replicated in an independent dataset. RESULTS Linked dimensions of psychopathology and connectivity We sought to delineate multivariate relationships between functional connectivity and psychiatric symptoms in a large sample of youth. To do this, we used sCCA, an unsupervised learning technique that seeks to find correlations between two high-dimensional datasets33. In total, we studied 999 participants of ages 8–22 who completed both functional neuroimaging and a comprehensive evaluation of psychiatric symptoms as part of the PNC27,32 (Table 1 and Fig. 1). Participants in the PNC were recruited from Children’s Hospital of Philadelphia pediatric network in the greater Philadelphia area. In this community-based study, participants were not recruited from psychiatric services. As such, the prevalence of screening into specific psychopathology categories generally aligned with epidemiologically ascertained samples, as previously described27 (see Supplementary Table 1). We divided this sample into discovery (n = 663) and replication datasets (n = 336) that were matched on age, sex, race, and overall psychopathology (Fig. 1 and Supplementary Fig. 1). Following pre-processing using a validated pipeline that minimizes the impact of in-scanner motion38 (see Supplementary Fig. 2, Supplementary Fig. 3 and Supplementary Fig. 4), we constructed subject-level functional networks using a 264-node parcellation system that includes an a priori assignment of nodes to network communities10 (Fig. 2a–c, e.g., modules or sub-networks; see Online Methods section). Prior to analysis with sCCA, we regressed age, sex, race, and motion out of both the connectivity and clinical data to ensure that these potential confounders did