Brain-Based Predictions of Psychiatric Illness-Linked Behaviors Across the Sexes

Sex-Differential Markers of Psychiatric Risk and Treatment Response Based on Premature Aging of Functional Brain Network Dynamics and Peripheral Physiology

BACKGROUND

Aging is a multilevel process of gradual decline that predicts morbidity and mortality. Independent investigations have implicated senescence of brain and peripheral physiology in psychiatric risk, but it is unclear whether these effects stem from unique or shared mechanisms.

METHODS

To address this question, we analyzed clinical, blood chemistry, and resting-state functional neuroimaging data in a healthy aging cohort (n = 427; ages 36-100 years) and 2 disorder-specific samples including patients with early psychosis (100 patients, 16-35 years) and major depressive disorder (MDD) (104 patients, 20-76 years).

RESULTS

We identified sex-dependent coupling between blood chemistry markers of metabolic senescence (i.e., homeostatic dysregulation), functional brain network aging, and psychiatric risk. In females, premature aging of frontoparietal and somatomotor networks was linked to greater homeostatic dysregulation. It also predicted the severity and treatment resistance of mood symptoms (depression/anxiety [all 3 samples], anhedonia [MDD]) and social withdrawal/behavioral inhibition (avoidant personality disorder [healthy aging], negative symptoms [early psychosis]). In males, premature aging of the default mode, cingulo-opercular, and visual networks was linked to reduced homeostatic dysregulation and predicted the severity and treatment resistance of symptoms relevant to hostility/aggression (antisocial personality disorder [healthy aging], mania/positive symptoms [early psychosis]), impaired thought processes (early psychosis, MDD), and somatic problems (healthy aging, MDD).

CONCLUSIONS

Our findings identify sexually dimorphic relationships between brain dynamics, peripheral physiology, and risk for psychiatric illness, suggesting that the specificity of putative risk biomarkers and precision therapeutics may be improved by considering sex and other relevant personal characteristics.

Reproducible Effects of Sex and Acquisition Order on Multiple Global Signal Metrics: Implications for Functional Connectivity Studies of Phenotypic Individual Differences Using fMRI

PURPOSE

The identification of relationships between individual differences in functional connectivity (FC) and behavior has been the focus of considerable investigation. Although emerging evidence has identified relationships between FC and cognitive performance, relationships between FC and measures of affect, including depressed mood, anhedonia, and anxiety, and decision-making style, including impulsivity and sensation seeking, appear to be more inconsistent across the literature. This may be due to low power, methodological differences across studies, including the use of global signal correction (GSR), or uncontrolled characteristics of the population.

METHODS

Here, we evaluated measures of FC, regional variance, and global signal (GS) across six functional MRI (fMRI) sequences of different tasks and resting states and their relationship with individual differences in self-reported measures of symptoms of depression, anxiety, impulsivity, reward sensitivity, and sensation seeking, as well as demographic variables and acquisition order, within groups of distressed and healthy young adults (18-25 years old).

FINDINGS

Adopting a training/testing sample structure to the analysis, we found no evidence of reproducible brain/behavior relationships despite identifying regions and connections that reflect reliable between-scan individual differences. However, summary measures of the GS were reproducibly associated with sex: The most consistent finding was an increase in low frequency variance of the blood-oxygenation-level-dependent (BOLD) signal from all gray matter regions in males relative to females. Post hoc analysis of GS topography yielded sex differences in a number of regions, including cerebellum and putamen. In addition, effects of paradigm acquisition order were observed on GS measures, including an increase in BOLD signal variance across time. In an exploratory analysis, a specific relationship between sex and relative high-frequency within-scanner motion was observed.

CONCLUSIONS

Together, the findings suggest that FC relationships with affective measures may be inconsistent or modest, but that global phenomena related to state and individual differences can be robust and must be evaluated, particularly in studies of psychiatric disorders such as mood disorders or ADHD, which show sex differences.

Considering the interconnected nature of social identities in neuroimaging research

Considerable heterogeneity exists in the expression of complex human behaviors across the cognitive, personality and mental health domains. It is increasingly evident that individual variability in behavioral expression is substantially affected by sociodemographic factors that often interact with life experiences. Here, we formally address the urgent need to incorporate intersectional identities in neuroimaging studies of behavior, with a focus on research in mental health. We highlight how diverse sociodemographic factors influence the study of psychiatric conditions, focusing on how interactions between those factors might contribute to brain biology and illness expression, including prevalence, symptom burden, help seeking, treatment response and tolerance, and relapse and remission. We conclude with a discussion of the considerations and actionable items related to participant recruitment, data acquisition and data analysis to facilitate the inclusion and incorporation of diverse intersectional identities in neuroimaging.

Mapping the neurodevelopmental predictors of psychopathology

Neuroimaging research has uncovered a multitude of neural abnormalities associated with psychopathology, but few prediction-based studies have been conducted during adolescence, and even fewer used neurobiological features that were extracted across multiple neuroimaging modalities. This gap in the literature is critical, as deriving accurate brain-based models of psychopathology is an essential step towards understanding key neural mechanisms and identifying high-risk individuals. As such, we trained adaptive tree-boosting algorithms on multimodal neuroimaging features from the Lifespan Human Connectome Developmental (HCP-D) sample that contained 956 participants between the ages of 8 to 22 years old. Our feature space consisted of 1037 anatomical, 1090 functional, and 192 diffusion MRI features, which were used to derive models that separately predicted internalizing symptoms, externalizing symptoms, and the general psychopathology factor. We found that multimodal models were the most accurate, but all brain-based models of psychopathology yielded out-of-sample predictions that were weakly correlated with actual symptoms (r2 < 0.15). White matter microstructural properties, including orientation dispersion indices and intracellular volume fractions, were the most predictive of general psychopathology, followed by cortical thickness and functional connectivity. Spatially, the most predictive features of general psychopathology were primarily localized within the default mode and dorsal attention networks. These results were mostly consistent across all dimensions of psychopathology, except orientation dispersion indices and the default mode network were not as heavily weighted in the prediction of internalizing and externalizing symptoms. Taken with prior literature, it appears that neurobiological features are an important part of the equation for predicting psychopathology but relying exclusively on neural markers is clearly not sufficient, especially among adolescent samples with subclinical symptoms. Consequently, risk factor models of psychopathology may benefit from incorporating additional sources of information that have also been shown to explain individual differences, such as psychosocial factors, environmental stressors, and genetic vulnerabilities.

The cell-type underpinnings of the human functional cortical connectome

The functional properties of the human brain arise, in part, from the vast assortment of cell types that pattern the cerebral cortex. The cortical sheet can be broadly divided into distinct networks, which are embedded into processing streams, or gradients, that extend from unimodal systems through higher-order association territories. Here using microarray data from the Allen Human Brain Atlas and single-nucleus RNA-sequencing data from multiple cortical territories, we demonstrate that cell-type distributions are spatially coupled to the functional organization of cortex, as estimated through functional magnetic resonance imaging. Differentially enriched cells follow the spatial topography of both functional gradients and associated large-scale networks. Distinct cellular fingerprints were evident across networks, and a classifier trained on postmortem cell-type distributions was able to predict the functional network allegiance of cortical tissue samples. These data indicate that the in vivo organization of the cortical sheet is reflected in the spatial variability of its cellular composition.

Brain-phenotype predictions of language and executive function can survive across diverse real-world data: Dataset shifts in developmental populations

Predictive modeling potentially increases the reproducibility and generalizability of neuroimaging brain-phenotype associations. Yet, the evaluation of a model in another dataset is underutilized. Among studies that undertake external validation, there is a notable lack of attention to generalization across dataset-specific idiosyncrasies (i.e., dataset shifts). Research settings, by design, remove the between-site variations that real-world and, eventually, clinical applications demand. Here, we rigorously test the ability of a range of predictive models to generalize across three diverse, unharmonized developmental samples: the Philadelphia Neurodevelopmental Cohort (n=1291), the Healthy Brain Network (n=1110), and the Human Connectome Project in Development (n=428). These datasets have high inter-dataset heterogeneity, encompassing substantial variations in age distribution, sex, racial and ethnic minority representation, recruitment geography, clinical symptom burdens, fMRI tasks, sequences, and behavioral measures. Through advanced methodological approaches, we demonstrate that reproducible and generalizable brain-behavior associations can be realized across diverse dataset features. Results indicate the potential of functional connectome-based predictive models to be robust despite substantial inter-dataset variability. Notably, for the HCPD and HBN datasets, the best predictions were not from training and testing in the same dataset (i.e., cross-validation) but across datasets. This result suggests that training on diverse data may improve prediction in specific cases. Overall, this work provides a critical foundation for future work evaluating the generalizability of brain-phenotype associations in real-world scenarios and clinical settings.

Influences of sex and gender on the associations between risk and protective factors, brain, and behavior

Risk and protective factors for psychiatric illnesses are linked to distinct structural and functional changes in the brain. Further, the prevalence of these factors varies across sexes and genders, yet the distinct and joint effects of sex and gender in this context have not been extensively characterized. This suggests that risk and protective factors may map onto the brain and uniquely influence individuals across sexes and genders. Here, we review how specific risk (childhood maltreatment, the COVID-19 pandemic, experiences of racism), and protective factors (social support and psychological resilience) distinctly influence the brain across sexes and genders. We also discuss the role of sex and gender in the compounding effects of risk factors and in the interdependent influences of risk and protective factors. As such, we call on researchers to consider sex and gender when researching risk and protective factors for psychiatric illnesses, and we provide concrete recommendations on how to account for them in future research. Considering protective factors alongside risk factors in research and acknowledging sex and gender differences will enable us to establish sex- and gender-specific brain-behavior relationships. This will subsequently inform the development of targeted prevention and intervention strategies for psychiatric illnesses, which have been lacking. To achieve sex and gender equality in mental health, acknowledging and researching potential differences will lead to a better understanding of men and women, males and females, and the factors that make them more vulnerable or resilient to psychopathology.

Generalizable and replicable brain-based predictions of cognitive functioning across common psychiatric illness

A primary aim of computational psychiatry is to establish predictive models linking individual differences in brain functioning with symptoms. In particular, cognitive impairments are transdiagnostic, treatment resistant, and associated with poor outcomes. Recent work suggests that thousands of participants may be necessary for the accurate and reliable prediction of cognition, questioning the utility of most patient collection efforts. Here, using a transfer learning framework, we train a model on functional neuroimaging data from the UK Biobank to predict cognitive functioning in three transdiagnostic samples (ns = 101 to 224). We demonstrate prediction performance in all three samples comparable to that reported in larger prediction studies and a boost of up to 116% relative to classical models trained directly in the smaller samples. Critically, the model generalizes across datasets, maintaining performance when trained and tested across independent samples. This work establishes that predictive models derived in large population-level datasets can boost the prediction of cognition across clinical studies.

Sex differences in deep brain shape and asymmetry persist across schizophrenia and healthy individuals: A meta-analysis from the ENIGMA-Schizophrenia Working Group

Background

Schizophrenia (SCZ) is characterized by a disconnect from reality that manifests as various clinical and cognitive symptoms, and persistent neurobiological abnormalities. Sex-related differences in clinical presentation imply separate brain substrates. The present study characterized deep brain morphology using shape features to understand the independent effects of diagnosis and sex on the brain, and to determine whether the neurobiology of schizophrenia varies as a function of sex.

Methods

This study analyzed multi-site archival data from 1,871 male (M) and 955 female (F) participants with SCZ, and 2,158 male and 1,877 female healthy controls (CON) from twenty-three cross-sectional samples from the ENIGMA Schizophrenia Workgroup. Harmonized shape analysis protocols were applied to each site's data for seven deep brain regions obtained from T1-weighted structural MRI scans. Effect sizes were calculated for the following main contrasts: 1) Sex effects;2) Diagnosis-by-Sex interaction; 3) within sex tests of diagnosis; 4) within diagnosis tests of sex differences. Meta-regression models between brain structure and clinical variables were also computed separately in men and women with schizophrenia.

Results

Mass univariate meta-analyses revealed more concave-than-convex shape differences in all regions for women relative to men, across diagnostic groups ( d = -0.35 to 0.20, SE = 0.02 to 0.07); there were no significant diagnosis-by-sex interaction effects. Within men and women separately, we identified more-concave-than-convex shape differences for the hippocampus, amygdala, accumbens, and thalamus, with more-convex-than-concave differences in the putamen and pallidum in SCZ ( d = -0.30 to 0.30, SE = 0.03 to 0.10). Within CON and SZ separately, we found more-concave-than-convex shape differences in the thalamus, pallidum, putamen, and amygdala among females compared to males, with mixed findings in the hippocampus and caudate ( d = -0.30 to 0.20, SE = 0.03 to 0.09). Meta-regression models revealed similarly small, but significant relationships, with medication and positive symptoms in both SCZ-M and SCZ-F.

Conclusions

Sex-specific variation is an overriding feature of deep brain shape regardless of disease status, underscoring persistent patterns of sex differences observed both within and across diagnostic categories, and highlighting the importance of including it as a critical variable in studies of neurobiology. Future work should continue to explore these dimensions independently to determine whether these patterns of brain morphology extend to other aspects of neurobiology in schizophrenia, potentially uncovering broader implications for diagnosis and treatment.

Key Points

Statistical analyses revealed significant main effects for diagnosis and sex in deep brain shape morphology. Among patients with schizophrenia, there was a pattern of thinning and surface contraction in the bilateral hippocampus, amygdala, accumbens, and thalamus, and a pattern of significant thickening and surface expansion in the bilateral putamen and pallidum compared to healthy control participants. Between males and females, there was a pattern of significant thinning and surface contraction in the bilateral thalamus, pallidum, putamen, and amygdala in females compared to males.There was no significant interaction between diagnosis and biological sex, suggesting that sex differences in deep brain shape and asymmetry among patients with schizophrenia reflect those observed in healthy individuals.Small but statistically significant relationships exist between brain structure and clinical correlates of schizophrenia were similar for both men and women with the disease, such that higher CPZ was associated with shape-derived thinning and surface contraction in the caudate, accumbens, hippocampus, amygdala, and thalamus, and elevated positive symptoms were associated with shape-derived thinning and surface contraction in the bilateral caudate, right hippocampus, and right amygdala.

Functional brain networks are associated with both sex and gender in children

Sex and gender are associated with human behavior throughout the life span and across health and disease, but whether they are associated with similar or distinct neural phenotypes is unknown. Here, we demonstrate that, in children, sex and gender are uniquely reflected in the intrinsic functional connectivity of the brain. Somatomotor, visual, control, and limbic networks are preferentially associated with sex, while network correlates of gender are more distributed throughout the cortex. These results suggest that sex and gender are irreducible to one another not only in society but also in biology.

Sex differences in the functional network underpinnings of psychotic-like experiences in children

Psychotic-like experiences (PLEs) include a range of sub-threshold symptoms that resemble aspects of psychosis but do not necessarily indicate the presence of psychiatric illness. These experiences are highly prevalent in youth and are associated with developmental disruptions across social, academic, and emotional domains. While not all youth who report PLEs develop psychosis, many develop other psychiatric illnesses during adolescence and adulthood. As such, PLEs are theorized to represent early markers of poor mental health. Here, we characterized the similarities and differences in the neurobiological underpinnings of childhood PLEs across the sexes using a large sample from the ABCD Study (n=5,260), revealing sex-specific associations between functional networks connectivity and PLEs. We find that although the networks associated with PLEs overlap to some extent across the sexes, there are also crucial differences. In females, PLEs are associated with dispersed cortical and non-cortical connections, whereas in males, they are primarily associated with functional connections within limbic, temporal parietal, somato/motor, and visual networks. These results suggest that early transdiagnostic markers of psychopathology may be distinct across the sexes, further emphasizing the need to consider sex in psychiatric research as well as clinical practice.

Brain-phenotype predictions can survive across diverse real-world data

Recent work suggests that machine learning models predicting psychiatric treatment outcomes based on clinical data may fail when applied to unharmonized samples. Neuroimaging predictive models offer the opportunity to incorporate neurobiological information, which may be more robust to dataset shifts. Yet, among the minority of neuroimaging studies that undertake any form of external validation, there is a notable lack of attention to generalization across dataset-specific idiosyncrasies. Research settings, by design, remove the between-site variations that real-world and, eventually, clinical applications demand. Here, we rigorously test the ability of a range of predictive models to generalize across three diverse, unharmonized samples: the Philadelphia Neurodevelopmental Cohort (n=1291), the Healthy Brain Network (n=1110), and the Human Connectome Project in Development (n=428). These datasets have high inter-dataset heterogeneity, encompassing substantial variations in age distribution, sex, racial and ethnic minority representation, recruitment geography, clinical symptom burdens, fMRI tasks, sequences, and behavioral measures. We demonstrate that reproducible and generalizable brain-behavior associations can be realized across diverse dataset features with sample sizes in the hundreds. Results indicate the potential of functional connectivity-based predictive models to be robust despite substantial inter-dataset variability. Notably, for the HCPD and HBN datasets, the best predictions were not from training and testing in the same dataset (i.e., cross-validation) but across datasets. This result suggests that training on diverse data may improve prediction in specific cases. Overall, this work provides a critical foundation for future work evaluating the generalizability of neuroimaging predictive models in real-world scenarios and clinical settings.

Distinct brain network features predict internalizing and externalizing traits in children and adults

Internalizing and externalizing traits are two distinct classes of behaviors in psychiatry. However, whether shared or unique brain network features predict internalizing and externalizing behaviors in children and adults remain poorly understood. Using a sample of 2262 children from the Adolescent Brain Cognitive Development (ABCD) study and 752 adults from the Human Connectome Project (HCP), we show that network features predicting internalizing and externalizing behavior are, at least in part, dissociable in children, but not in adults. In ABCD children, traits within internalizing and externalizing behavioral categories are predicted by more similar network features concatenated across task and resting states than those between different categories. We did not observe this pattern in HCP adults. Distinct network features predict internalizing and externalizing behaviors in ABCD children and HCP adults. These data reveal shared and unique brain network features accounting for individual variation within broad internalizing and externalizing categories across developmental stages.

Aging brain shows joint declines in brain within-network connectivity and between-network connectivity: a large-sample study (N > 6,000)

Introduction

Numerous studies have shown that aging has important effects on specific functional networks of the brain and leads to brain functional connectivity decline. However, no studies have addressed the effect of aging at the whole-brain level by studying both brain functional networks (i.e., within-network connectivity) and their interaction (i.e., between-network connectivity) as well as their joint changes.

Methods

In this work, based on a large sample size of neuroimaging data including 6300 healthy adults aged between 49 and 73 years from the UK Biobank project, we first use our previously proposed priori-driven independent component analysis (ICA) method, called NeuroMark, to extract the whole-brain functional networks (FNs) and the functional network connectivity (FNC) matrix. Next, we perform a two-level statistical analysis method to identify robust aging-related changes in FNs and FNCs, respectively. Finally, we propose a combined approach to explore the synergistic and paradoxical changes between FNs and FNCs.

Results

Results showed that the enhanced FNCs mainly occur between different functional domains, involving the default mode and cognitive control networks, while the reduced FNCs come from not only between different domains but also within the same domain, primarily relating to the visual network, cognitive control network, and cerebellum. Aging also greatly affects the connectivity within FNs, and the increased within-network connectivity along with aging are mainly within the sensorimotor network, while the decreased within-network connectivity significantly involves the default mode network. More importantly, many significant joint changes between FNs and FNCs involve default mode and sub-cortical networks. Furthermore, most synergistic changes are present between the FNCs with reduced amplitude and their linked FNs, and most paradoxical changes are present in the FNCs with enhanced amplitude and their linked FNs.

Discussion

In summary, our study emphasizes the diversity of brain aging and provides new evidence via novel exploratory perspectives for non-pathological aging of the whole brain.