Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Sep 21.
Published in final edited form as: Psychiatry Res Neuroimaging. 2026 Jun 15;362:112274. doi: 10.1016/j.pscychresns.2026.112274

Brain Functional Connectivity Patterns Associated with Adiposity in Schizophrenia

Keith Dodd 1,2, Kristina T Legget 1,3, Marc-Andre Cornier 4, Maureen McHugo 1, Korey P Wylie 1, Andrew M Novick 1, Brian D Berman 5, Jason R Tregellas 1,3
PMCID: PMC13589839  NIHMSID: NIHMS2191174  PMID: 42322676

Abstract

Individuals with schizophrenia (SZ) experience significantly greater obesity rates compared to the general population. The underlying biological mechanisms, however, remain poorly understood. Although brain functional connectivity (FC) has been associated with obesity in the general population, its role in obesity in SZ is largely unexplored. As such, this study examined FC contributions to adiposity in participants with SZ (n=46) and an adiposity-matched comparison group without SZ (n=46) as they completed resting-state functional magnetic resonance imaging in fasted and fed states. A machine learning approach identified FC patterns associated with adiposity (percent body fat), with model performance evaluated on an independent test set. In both fasted and fed states, brain regions involved in reward and eating behaviors contributed to adiposity prediction in both groups. While fasted, predictive connections in SZ largely involved limbic and sensorimotor networks, whereas comparison group networks were more varied. In the fed state, SZ predictive features included visual, default mode, and executive control networks, while comparison group connections involved limbic, sensorimotor, salience, and executive control networks. Findings may suggest shared, as well as diagnosis- and state-specific, FC patterns associated with adiposity in SZ, which may help inform future development of obesity-related interventions in this high-risk population.

Keywords: schizophrenia, obesity, machine learning, functional connectivity, fMRI, fasted, fed

1. INTRODUCTION

Schizophrenia (SZ), a leading cause of disability worldwide, is a chronic psychiatric disorder that affects nearly 1% of the global population (McGrath et al., 2008). In addition to severe psychiatric symptoms, SZ is highly associated with obesity (Annamalai et al., 2017; Subashini et al., 2011), with an incidence three times greater than the general population (Chen et al., 2022). This burden significantly contributes to decreased quality of life and increased rates of morbidity and mortality experienced by individuals with SZ (Annamalai et al., 2017). Despite this, the mechanisms contributing to obesity in SZ are not fully understood. Recent neuroimaging studies, however, have demonstrated associations between obesity in SZ and altered brain structure and function (McWhinney et al., 2022), suggesting that neuronal function may contribute to this comorbidity.

Functional magnetic resonance imaging (fMRI) studies have increasingly linked obesity with disruptions in functional connectivity (FC) in the general population. Specifically, obesity has been associated with hypoconnectivity within the “triple network” (comprised of the default mode, executive control, and salience networks) (Ding et al., 2020; Guo et al., 2023; Lee et al., 2022; Park et al., 2020), as well as hyperconnectivity between the reward, limbic, and salience networks (Li et al., 2023; Ravichandran et al., 2021). Higher body mass index (BMI) has also been associated with greater temporal variability in FC (i.e., more fluctuations in strength or pattern of connectivity over time between reward/limbic and default mode, executive control, salience, and sensorimotor networks) suggesting altered integration of motivational signals with motor processes relevant to eating behavior (Guo et al., 2023). Such disrupted FC may contribute to obesity by modulating impulsivity, satiety perception, eating behaviors, and cognitive control (Parsons et al., 2022).

Notably, many of the same networks likely related to obesity in the general population appear to be associated with the neurobiology of SZ itself. For example, triple network dysconnectivity is thought to contribute to SZ symptoms (Aryutova and Stoyanov, 2021). Similarly, increased mesolimbic dopamine signaling in SZ is linked to aberrant reward and salience processing, while decreased mesocortical dopamine signaling likely contributes to the observed hypoconnectivity of limbic and reward networks with executive control and default mode networks (Richter et al., 2015; Robison et al., 2020; Simon et al., 2010; Wang et al., 2023; Zeng et al., 2022). Dysconnectivity between reward and sensorimotor/visual networks has also been linked to both dopaminergic system dysfunction and the auditory and visual hallucinations commonly experienced in SZ (Cao et al., 2022; Rolland et al., 2015).

Despite the similarities in FC alterations in SZ and obesity in the general population, little is known about how FC relates to obesity in schizophrenia specifically. As such, there is a need to better understand the FC patterns associated with obesity in SZ and how these relate to obesity in the general population. Importantly, identifying functional connectivity patterns associated with adiposity in SZ may help inform targeted treatment development for obesity in this vulnerable population. Such associations between FC and obesity in SZ, however, are likely complex. To help parse this complexity, the current study employed a machine learning model capable of capturing non-linear associative patterns between adiposity (percent body fat) and FC to allow for the detection of subtle, multivariate patterns that may not be captured by other approaches.

Given prior evidence that FC associated with obesity is modulated by satiety state (Li et al., 2023), both fasted and fed states were examined. Due to the overlapping nature of network alterations observed in both SZ and obesity, we hypothesized that FC primarily between the reward, limbic, and triple networks would relate to adiposity in both individuals with SZ and an adiposity-matched comparison group without SZ. As prior studies have linked SZ neurobiology and symptoms to dysconnectivity between reward and sensorimotor/visual networks (Cao et al., 2022; Rolland et al., 2015), we further anticipated that alterations in these connections would be associated with adiposity in those with SZ.

2. METHODS

2.1. Participant Characteristics and Recruitment

Participants were recruited from two larger studies targeting exercise and diet effects in individuals with overweight/obesity and schizophrenia (SZ; Study 1: NCT02455193) (Dodd et al., 2024) or those with overweight/obesity without SZ (Study 2: NCT02047721) (Cosgrove et al., 2025; Purcell et al., 2022). Participants provided written informed consent approved by the Colorado Multiple Institutional Review Board and were recruited via clinician referrals, community outreach, and advertisements. All participants were weight-stable (±5% for ≥3 months) and excluded for current pregnancy, significant endocrine/metabolic disease, illicit substance use, MRI contraindications, or bariatric surgery history. For participants with SZ, psychiatric medication type/dose were also stable for at least three months prior to study participation. Additional eligibility criteria specific to each study are detailed in Supplementary Material 1.1.

2.2. Study Design

The current work focuses on baseline measures from both studies. All procedures took place at the University of Colorado Anschutz Medical Campus (CU AMC). Prior to scan day, adiposity (percent body fat) was measured from a body composition assessment (Study 1: BOD POD, COSMED (Dempster and Aitkens, 1995), Study 2: dual-energy X-ray absorptiometry [DXA]; Discovery W, Hologic, Marlborough, MA). BOD POD and DXA measures have demonstrated agreement within approximately 1% body fat in adults (Fields et al., 2002). Participants also completed a one-day, macronutrient-controlled, eucaloric run-in diet the day before scanning. On scan day, after an overnight fast, participants underwent resting-state fMRI in the fasted state and then 30 minutes after a standardized breakfast meal (fed state scan). Additional run-in diet and breakfast meal details can be found in Supplementary Material 1.2. Nicotine (if applicable) and caffeine abstention (both ≥1 hr) were confirmed by research staff, as participants arrived for the study visit more than an hour prior to scanning. Following completion of both scans, participants with SZ also completed the Brief Psychiatric Rating Scale (BPRS) (Overall, 1962) and the Scale for the Assessment of Negative Symptoms (SANS) (Andreasen, 1989), to assess clinical symptoms. Other hormonal, behavioral, and cognitive outcomes were also collected in the larger studies but are outside the scope of the current investigation.

2.3. fMRI Data Acquisition

Imaging was performed using a Siemens Skyra 3.0 T with a 20-channel head coil. A T1-weighted magnetization-prepared rapid acquisition gradient-echo (MPRAGE) anatomical image was acquired for each participant. Resting-state functional images were then acquired with an echo planar T2* blood oxygen level dependent (BOLD) sequence (additional acquisition details provided in Supplementary Material 1.3).

2.4. Data Preprocessing

Data were preprocessed with fMRIPrep v20.2.6 (Esteban et al., 2018) (see Supplementary Material 1.4 for details), including slice timing correction, motion correction, and normalization to the Montreal Neurological Institute (MNI) 2009c template. Visual inspection for alignment of the normalized structural and functional data to the MNI template was then performed for each participant’s data.

2.5. Data Denoising and Quality Control

Denoising was conducted using the Comprehensive Independent Component Analysis Denoising Assistant (CICADA) v1.2 (https://github.com/keithcdodd/CICADA (Dodd et al., 2025)). Details on the methods and specifications used for CICADA can be found in Supplementary Material 1.5.

2.6. Data Analyses

2.6.1. Feature Extraction

The mean timeseries was extracted from all regions-of-interest (ROIs) in the Brainnetome atlas (Fan et al., 2016) for each participant and feeding state (fasted and fed). To avoid including areas with signal dropout, only the voxels present in all participants’ functional masks, as provided by CICADA, were included. Partial correlations between all ROI pairs were computed using precision matrix estimation (Wang et al., 2016) to generate connectivity matrices. All ROI pairs from the lower triangle of the connectivity matrices (i.e., all non-redundant pairwise connections excluding the diagonal) served as model features. Altogether, this resulted in 246 ROIs with 30,135 unique functional connections as initial features.

2.6.2. Data Splitting and Confound Regression

Feature sets were automatically stratified into six data subsets matched in age, sex, and adiposity (see Supplementary Material 1.10), allowing one subset (n = 8) to be reserved for independent testing and the remaining five subsets (n = 38 total) for training and optimization. This structure supports stratified five-fold cross-validation for hyperparameter tuning, in line with established practices (Dunias et al., 2024), and closely approximates the commonly adopted 80/20 training/testing split for model evaluation (Sivakumar et al., 2024). Age and sex effects were minimized by the cross-validated version of confound regression (Snoek et al., 2019), such that age and sex effects were estimated in the training data only, with their estimated linear coefficients then regressed out of the training and testing data. Model training and evaluation were then conducted on the training and testing sets, respectively.

2.6.3. Feature Selection and Model Training

To reduce redundancy, features with high inter-correlation (R2 > 0.5) were pruned by retaining a representative from each correlated group. Models were trained using least-squares boosting (LSBoost) with five-fold cross-validation and recursive feature elimination, first jointly and then separately by group. Features with low proportional importance (PI < 0.01; see Supplementary Material 1.7 for more details on PI) were iteratively removed, and correlated features previously dropped were reconsidered in the final expanded model (see Supplementary Material 1.6, 1.7, and Supplementary Figure 1 for more details).

2.6.4. Model Evaluation and Network Identification

Model performance was assessed using permutation testing of mean squared error (MSE) and on training and independent testing sets, and across groups (see Supplementary Material 1.7 and 1.8). To support network-level interpretations, ROIs were assigned to canonical resting-state networks (Yeo et al., 2011), with additional labeling of reward/eating-related regions (referred to as the reward and appetitive network [RAN] hereon) based on Brainnetome metadata (Fan et al., 2016). See Supplementary Material 1.9, 1.11, and 1.12 for more details on the identification of reward/eating-related regions.

To further assess sensitivity of model performance to individual participants within the independent test sets, leave-one-subject-out sensitivity analyses were performed by recalculating test-set Pearson correlations after sequential removal of each test-set participant. Model performance was additionally characterized using Pearson correlation (r) and coefficient of determination (r2) values between predicted and observed adiposity values (Supplementary Material Table 17).

2.6.5. Model Interpretation

More influential functional connections in predicting adiposity were characterized by increased PI values (see section 2.6.3). The overall directionality of the relationships between FC and adiposity was determined by examining partial correlations between the FC of each ROI-pair across participants and adiposity, controlling for age and sex. For visual clarity, only the top five functional connections were included in figures with more details given in Supplementary Tables 6, 7, and 9-12. Between-network predictive importance was estimated by summing PI values across all ROI pairs linking each unique network pair (“between-network summed importance”). Similarly, within-network predictive importance was estimated by summing PI values across all ROI pairs within each network (“within-network summed importance”). Tables including information on all retained features for each model can be found in the Supplementary Material (Supplementary Tables 9-12).

3. RESULTS

3.1. Participant Characteristics

Participant characteristics are presented in Table 1. Age and sex differences between groups were adjusted for by regressing out age and sex effects from the functional connectivity data (see Methods). Following exclusion due to missing data, retained noise (as indicated by CICADA), and adiposity matching between groups, 46 participants remained in each group. Full exclusion criteria and matching procedure are described in the Supplementary Material (1.1, 1.5, 1.10). The groups in the final analyses did not significantly differ in adiposity (Table 1).

Table 1.

Participant Characteristics

Schizophrenia (SZ) Adiposity-Matched Comparison Group (CG) Group Differences
Mean (SD) Mean (SD) p-value a
N 46 46 -
Age (yrs) 44.9 (13.9) 37.3 (9.9) 0.003
Sex(F/M) 21/25 38/8 <0.001
Adiposity (%bf) 39.1 (9.1) 40.0 (6.7) 0.629
BMIb (kg/m2) 31.0 (5.9) 30.6 (3.2) 0.709
Nicotine Use (Y/N) 25/21 0/46 <0.001
BPRS: Positive Symptomsb 6.2 (4.0) - -
BPRS: Negative Symptomsb 5.1 (3.0) - -
SANS: Total 26.0 (21.6) - -
Chlorpromazine Equivalentsb (mg) 354.3 (337.3) - -
Median FD (mm) 0.29 (0.16) 0.22 (0.10) <0.001
Open in a new tab
a

Two-tailed t-tests were used to determine group differences, except for sex and nicotine, for which chi-squared tests were used. Significant p-values (<0.05) are bolded.

b

Sample size for the following measures differed due to missing data (sample size for each group as follows: comparison group, SZ): BMI (38, 46), BPRS Positive Symptoms (N/A, 45), BPRS Negative Symptoms (N/A, 45), chlorpromazine equivalents (N/A, 42).

Abbreviations: SD: standard deviation; yrs: years; F/M: female/male; %bf: percent body fat; Y/N: yes/no; BPRS: Brief Psychiatric Rating Scale; SANS: Scales for the Assessment of Negative Symptoms; Eq: Equivalent; FD: framewise displacement.

Comparison of participant characteristics for the training and testing sets are displayed in Supplementary Table 3. For both the SZ and comparison groups, the testing set did not significantly differ from the training set in age, sex, adiposity, or median framewise displacement.

3.2. Fasted-State Results

3.2.1. Fasted-State Model Accuracy

Within groups, the final fasted-state models (Figure 1) were highly accurate for the cross-validated training data (SZ: mean absolute error (MAE)Training = 2.79% body fat, p < 0.001; comparison group MAETraining = 2.20% body fat, p < 0.001). Within-group accuracy of the fasted-state SZ and comparison models were similar in the testing sets (SZ MAETesting = 5.28% body fat; comparison group MAETesting = 5.31% body fat; p = 0.958), with accuracy nearly significant in the SZ testing data (SZ p = 0.056; comparison group p = 0.289). Across groups in the fasted state, the SZ model predicted adiposity in the comparison group more accurately than the comparison model predicted adiposity in the SZ group (SZ-model/comparison-group MAETesting = 4.69% body fat; comparison-model/SZ-group MAETesting = 9.08% body fat; p = 0.042). Within-group accuracy (i.e., the accuracy of a model trained and tested on data from the same group) was not significantly different from across-group accuracy (i.e., the accuracy of a model trained on one group but tested on the other) for either the comparison group (p = 0.485) or the SZ group (p = 0.137). Neither fasted-state model demonstrated significant accuracy bias related to age, sex, or median framewise displacement, as indicated by non-significant correlations between model residuals and these variables in the testing data (Supplementary Table 4).

Figure 1. Fasted-State Model Accuracy.

Figure 1.

Open in a new tab

(A) Within-Group Training Accuracy: five-fold cross-validated model fitting on the fasted-state training data; (B) Within-Group Testing Accuracy: model fitting on the corresponding independent fasted-state testing data; (C) Across-Group Testing Accuracy: model fitting on independent fasted-state testing data across groups. The dashed black line is a reference line of perfect predictions. Abbreviations: SZ: schizophrenia; CG: adiposity-matched comparison group; MAE: mean absolute error of adiposity predictions to true adiposity (% body fat).

3.2.2. Fasted-State Functional Connections Associated with Adiposity

In the comparison group, multiple brain networks, including the visual, dorsal attention, executive control, salience, limbic, and default mode networks, contributed to adiposity prediction (Table 2; Figure 3). Despite this diversity of network involvement, 50% of the associated regions in the ten most predictive connections were involved in the reward and appetitive network (Supplementary Table 6; Figure 2). For the SZ group, connectivity most predictive of adiposity largely included internetwork limbic connectivity (Table 2; Figure 3), especially between the limbic and sensorimotor networks (PI = 0.44). Salience and executive control network connectivity was also particularly predictive in the SZ group (PI = 0.15; Table 2). Of the associated regions, 35% of the ten most predictive connections were involved in the reward and appetitive network (Supplementary Table 6). Out of the top ten predictive connections, there were no regions in common between the SZ and comparison groups in the fasted state.

Table 2.

Network Connectivity Associated with Adiposity in the Fasted State

Schizophrenia (SZ)
Network 1 Network 2 Summed Importance
Limbic Sensorimotor 0.44
Executive Control Salience 0.15
Limbic Salience 0.14
Default Mode Limbic 0.08
Dorsal Attention Limbic 0.06
Executive Control Limbic 0.05
Limbic Visual 0.05
Default Mode Dorsal Attention 0.01
Limbic Limbic 0.01
Sensorimotor Sensorimotor 0.01
Adiposity-Matched Comparison Group (CG)
Network 1 Network 2 Summed Importance
Visual Visual 0.21
Dorsal Attention Executive Control 0.16
Salience Visual 0.15
Dorsal Attention Limbic 0.14
Default Visual 0.10
Executive Control Executive Control 0.10
Default Limbic 0.06
Salience Sensorimotor 0.04
Executive Control Sensorimotor 0.01
Executive Control Visual 0.01
Open in a new tab

Table 2 displays the 10 network connections most predictive of adiposity in the fasted state for each group (SZ & CG). Network connections’ summed importances were calculated by summating the importances of the given network connections from Supplementary Table 6 from each associated region. This same information is displayed in visual form in Figure 3.

Figure 3. Fasted-State Network Connectivity Associated with Adiposity.

Figure 3.

Open in a new tab

Network connectivity predictive of adiposity for the individuals with schizophrenia (SZ; left) and the adiposity-matched comparison group (CG; right) in the fasted state. Node size and line width reflect within- and between-network connectivity strength, respectively, with larger sizes indicating stronger connectivity.

Figure 2. Fasted-State Functional Connections Associated with Adiposity in Individuals with Schizophrenia (SZ) and the Adiposity-Matched Comparison Group (CG).

Figure 2.

Open in a new tab

The five most important predictors of adiposity for the individuals with schizophrenia (SZ: top) and the adiposity-matched comparison group (CG; bottom) in the fasted state. Edge width denotes relative predictor importance value. All region names are as depicted in Brainnetome (see Supplementary Material 1.11).

3.3. Fed-State Results

3.3.1. Fed-State Model Accuracy

Within groups, the final fed-state models (Figure 4) were highly accurate for the cross-validated training data (SZ: MAETraining = 3.79% body fat, p < 0.001; comparison group MAETraining = 3.62% body fat, p < 0.001). Within-group accuracy of the fed-state SZ and comparison models were similar in the testing sets (SZ MAETesting = 5.12% body fat; comparison group MAETesting = 4.57% body fat, p = 0.924) with significant accuracy in the SZ testing data (SZ p = 0.012; comparison group p = 0.161). No difference was observed in model accuracy across groups (SZ-model/comparison-group MAETesting = 5.04% body fat; comparison-model/SZ-group MAETesting = 7.61% body fat; p = 0.349). Within-group accuracy was not significantly different from across-group accuracy for either the SZ group (p = 0.150) or the comparison group (p = 0.689). Only the fed-state comparison model demonstrated significant accuracy bias, in sex (p = 0.006) but not in age nor median framewise displacement in the testing data (Supplementary Table 5). Leave-one-subject-out sensitivity analyses and corresponding test-set Pearson correlation (r) and coefficient of determination (r2) values for all independent testing analyses are reported in Supplementary Table 17.

Figure 4. Fed-State Model Accuracy.

Figure 4.

Open in a new tab

(A) Within-Group Training Accuracy: five-fold cross-validated model fitting on the fed-state training data; (B) Within-Group Testing Accuracy: model fitting on the corresponding independent fed-state testing data; (C) Across-Group Testing Accuracy: model fitting on independent fed-state testing data across groups. The dashed black line is a reference line of perfect predictions. Abbreviations: CG: adiposity-matched comparison group; SZ: schizophrenia; MAE: mean absolute error of adiposity predictions to true adiposity (% body fat).

3.3.2. Fed-State Connections Associated with Adiposity

Similar to the SZ group in the fasted state, comparison group connections most predictive of adiposity in the fed state included limbic and sensorimotor connectivity (PI = 0.35) as well as salience and executive control connectivity (PI = 0.21; Figure 6; Table 3). In contrast, SZ connections most predictive of adiposity in the fed state were dominated by connectivity within the visual network (PI = 0.55) and between the default mode and executive control networks (PI = 0.13; Figure 6; Table 3). Among the top ten connections, 25% and 30% were involved in the reward and appetitive network for the SZ and comparison groups, respectively (Supplementary Table 7; Figure 5). Unlike in the fasted state, there were a number of shared regions predictive of adiposity in the fed state for both groups, including the occipital pole, rostroposterior superior temporal sulcus, and the upper limb region of the primary motor cortex. Cross-satiety comparisons revealed distinct but partially overlapping sets of predictive regions across groups (see Supplementary Tables 6, 7, 9-12).

Figure 6. Fed-State Network Connectivity Associated with Adiposity.

Figure 6.

Open in a new tab

Network connectivity predictive of adiposity for the individuals with schizophrenia (SZ; left) and the adiposity-matched comparison group (CG; right) in the fed state. Node size and line width reflect within- and between-network connectivity strength, respectively, with larger sizes indicating stronger connectivity.

Table 3.

Network Connections Associated with Adiposity in the Fed State

Schizophrenia (SZ)
Network 1 Network 2 Summed Importance
Visual Visual 0.55
Default Executive Control 0.13
Executive Control Limbic 0.05
Dorsal Attention Sensorimotor 0.04
Dorsal Attention Dorsal Attention 0.04
Sensorimotor Visual 0.04
Salience Visual 0.04
Executive Control Salience 0.02
Dorsal Attention Limbic 0.02
Default Limbic 0.02
Adiposity-Matched Comparison Group (CG)
Network 1 Network 2 Summed Importance
Limbic Sensorimotor 0.35
Executive Control Salience 0.21
Executive Control Limbic 0.07
Sensorimotor Sensorimotor 0.06
Sensorimotor Visual 0.05
Default Mode Executive Control 0.04
Salience Visual 0.04
Limbic Salience 0.03
Salience Sensorimotor 0.03
Salience Salience 0.03
Open in a new tab

Table 3 displays the 10 network connections most predictive of adiposity in the fed state for each group (SZ & comparison group). Network connections’ summed importances were calculated by summating the importances of the given network connections from Supplementary Table 7 from each associated region. This same information is displayed in visual form in Figure 6.

Figure 5. Fed-State Functional Connections Associated with Adiposity in Individuals with SZ and the Adiposity-Matched Comparison Group (CG).

Figure 5.

Open in a new tab

The five most important predictors of adiposity for the individuals with schizophrenia (SZ; top), and the adiposity-matched comparison group (comparison group; bottom) in the fed state. Edge width denotes relative predictor importance value. All region names are as depicted in Brainnetome (see Supplementary Material 1.11).

4. DISCUSSION

This study aimed to identify resting-state functional connectivity (FC) patterns associated with adiposity in SZ utilizing machine learning-based predictive modeling. Both shared and unique resting-state FC patterns associated with adiposity were examined in individuals with SZ and adiposity-matched comparisons. As hypothesized, FC within and between the reward, limbic, and triple networks (salience, executive control, and default mode) were associated with adiposity in both groups. Beyond these shared associations, FC between the limbic and sensorimotor networks while fasted, and within the visual network while fed, appeared more strongly associated with adiposity in SZ, consistent with prior evidence linking SZ-related dopaminergic dysfunction to dysconnectivity between reward, sensorimotor, and visual networks (Cao et al., 2022; Rolland et al., 2015). To our knowledge, this is the first study to examine resting-state FC predictive of obesity across diagnostic groups (SZ or an adiposity-matched comparison group) and the first to investigate this across satiety states (fasted or fed). Together, these findings highlight: (1) common neural correlates of adiposity across SZ and non-SZ groups involving reward, limbic, and triple networks; (2) SZ-specific associations with limbic–sensorimotor and visual connectivity; and (3) modulation of these associations by satiety state, which may suggest potential state-dependent network correlates relevant to obesity interventions in SZ.

Across all conditions, FC predictive of adiposity involved regions strongly associated with food- and reward-related behavior, including the nucleus accumbens, hippocampus, amygdala, insula, and caudate (Donofry et al., 2020; Parsons et al., 2022). The consistent involvement of such regions aligns with the central role of dopaminergic reward circuits, primarily the mesolimbic pathway, in eating behavior and reward signaling (Volkow et al., 2011). As SZ is characterized by altered dopamine signaling, including in the mesolimbic pathway (Meisenzahl et al., 2007), this may suggest a neurobiological intersection between SZ pathophysiology and obesity. These findings suggest that reward-related circuits may reflect a transdiagnostic neural correlate associated with adiposity, with potential relevance in SZ, where mesolimbic dysregulation and antipsychotic effects may further modify obesity risk.

In the fasted state, the involvement of reward-and eating-related regions across numerous brain networks in predicting adiposity in the comparison group may reflect integration of sensory, attentional, and cognitive processes in food-seeking behavior while hungry (Carnell et al., 2012; Donofry et al., 2020; Parsons et al., 2022). In the SZ group, the dominance of limbic network interconnectivity importance in the fasted state may suggest that emotional regulation of reward evaluation plays a more central role in hunger management, aligning with mesolimbic dopamine dysregulation and altered food cue responsivity in SZ (Grimm et al., 2017; Meisenzahl et al., 2007). Similarly, the relatively high predictive importance (i.e., ranking among the strongest contributors within the given model) of limbic and sensorimotor connectivity in the SZ group when fasted may relate to heightened anticipation of food consumption (Contreras-Rodríguez et al., 2017), consistent with increased sensorimotor responsivity to food cues in individuals with SZ treated with high weight-gain risk APs (Stip et al., 2012). The involvement of salience and executive control network connectivity may reflect impaired cognitive control of salient food processing in SZ in the fasted state (Borowitz et al., 2020). While both groups demonstrated reward-related involvement, the lack of overlapping predictive regions may support distinct fasted-state FC patterns relating to adiposity in individuals with and without SZ.

In the fed state for the comparison group, the importance of limbic-sensorimotor and salience-executive control networks in predicting adiposity may reflect obesity-related alterations in eating anticipation, postprandial control, and interoceptive awareness (Borowitz et al., 2020; Contreras-Rodríguez et al., 2017; García-García et al., 2013; Guo et al., 2023; Gupta et al., 2020). Regions predictive of adiposity overlapped between the comparison group in the fed state and the SZ group in the fasted state, including regions implicated in reward, anticipation, and interoception (i.e., posterior cingulate cortex, anterior insula, and thalamus) (Leenaerts et al., 2022; Millan et al., 2017; Simmons and DeVille, 2017). While this overlap could suggest that similar neural substrates may contribute to obesity across conditions, closer inspection of the underlying connectivity patterns indicates that these shared regions may participate in distinct network interactions. For example, a posterior cingulate region (the caudal aspect of Brodmann area 23), involved in default mode and reward processing (Andrews-Hanna, 2012), was predictive of adiposity in both the comparison group in the fed state and the SZ group in the fasted state. In the comparison group while fed, this region was connected to salience-related circuitry with a positive association with adiposity, whereas in the SZ group while fasted, this region was functionally connected to executive control circuitry with a negative association with adiposity. Altogether, these patterns may suggest that while overlapping regions may be implicated across groups and satiety states, the direction and functional context of their network interactions may differ, potentially reflecting increased salience-driven reward processing in the comparison group post-prandially and decreased cognitive control over reward evaluation in SZ while fasted.

For the SZ group in the fed state, the strong influence of the visual network in predicting adiposity may relate to heightened internal food-related visual imagery post-meal, or alternatively, altered visual processing related to food cues (Avery et al., 2025, 2021). The involvement of connectivity between the default mode and executive control networks may reflect impaired inhibitory control of internal motivation for food reward despite satiety (Ding et al., 2020). The increase in the number of shared regions predictive of adiposity in the fed state, as compared to the fasted state, across groups may suggest partial convergence of adiposity-related FC patterns across groups postprandially. Similar to the comparison between the fasted-state SZ group and the fed-state comparison group, differences in the directionality of predictions, in terms of increases or decreases in adiposity, and connected regions may alternatively suggest that the shared regions still reflect distinct underlying neuronal mechanisms.

Within the comparison group, involvement of the rostral hippocampus and caudal aspect of Brodmann area 45 (i.e., inferior frontal gyrus) in both fasted and fed states may suggest that inhibitory control and food-related memory play a consistent role in obesity, regardless of satiety. Indeed, Brodmann area 45 has been associated with inhibitory control of appetite (Ishida et al., 2022; Takada et al., 2018), with the rostral aspect of the hippocampus being involved in memory-related aspects of eating behavior, gustation, and reward (Kanoski and Grill, 2017). Within the SZ group, consistent involvement of rostral aspects of Brodmann’s areas 35 and 36 (parahippocampal gyrus), the lateral amygdala, and the caudal aspect of Brodmann area 39 (angular gyrus) in both feeding states may suggest persistence of fasted-like FC patterns in individuals with SZ postprandially. This is consistent with prior associations between motivational food-seeking behavior while fasted (parahippocampal gyrus, amygdala) (Brooks et al., 2013; Burgess et al., 2016; Morris and Dolan, 2001) and modulations of neuronal response to sweet stimuli associated with BMI (angular gyrus) (Chen and Zeffiro, 2020). Notably, convergent evidence indicates that hippocampal hyperactivity and broader hippocampal–limbic dysfunction are core features of schizophrenia and are linked to dopaminergic and salience-processing abnormalities (Grace, 2012; Kätzel et al., 2020; Lodge and Grace, 2011; Tregellas et al., 2014), providing additional context for the prominence of medial temporal (e.g., parahippocampal, hippocampal) connectivity observed here.

The similar performance of across-group predictions as compared to within-group predictions is suggestive of considerable overlap in resting-state FC patterns predictive of adiposity in individuals with and without SZ. In the fasted state, applying the comparison group model to the SZ group resulted in greater prediction error than applying the SZ group model to the comparison group, which may reflect greater complexity in obesity-related FC patterns in SZ compared to the general population. Rather than reflecting distinct or uniform mechanisms, this finding may stem from the substantial functional dysconnectivity observed in SZ (Dong et al., 2018; Li et al., 2019): while obesity in SZ may involve many of the same neuronal processes as in the general population, the widespread network disruptions observed in individuals with SZ may give rise to more intricate FC interactions contributing to obesity.

Overall, although preliminary, the current findings may have important clinical implications for treating obesity in SZ. Strategies targeting reward-related and limbic network function, including approaches that alter activity within reward circuitry or associated neuromodulatory systems, may prove beneficial, especially while fasted. Postprandially, treatments aimed at reducing visual food cue reactivity or modulating default mode and salience network engagement may also be useful. While such network-level targets may be relevant to obesity more broadly, the distinct neural patterns observed in SZ may suggest that obesity interventions may need to be tailored to account for diagnosis-specific neural mechanisms and treatment contexts. These findings should be interpreted cautiously, however, as resting-state FC associations do not permit direct mechanistic inference, and similar levels of adiposity may arise through multiple interacting behavioral, environmental, metabolic, and neural pathways. Additionally, as all participants had overweight/obesity, the present findings reflect variability within elevated adiposity ranges rather than differences in comparison to normative-weight physiology.

Strengths of the current study include the use of machine learning techniques that capture nonlinear patterns, inclusion of both fasted and fed conditions, comparison to an adiposity-matched group, and implementation of independent testing and permutation-based evaluation procedures. Findings should, however, be considered in the context of study limitations. Generalizability is limited by the sample size, high dimensionality of the FC feature space, the modest independent test set size, and by group differences in sex and age. To help mitigate the impact of the sex and age differences, we regressed out linear effects of age and sex from the FC data prior to model training and evaluated this strategy by testing for prediction biases related to age and sex in the testing data. Only the fed-state comparison group model showed a significant bias, specifically in relation to sex, indicating differential prediction performance between males and females. Thus, while future work should replicate these findings in larger, age- and sex-matched samples and directly examine how these variables influence adiposity-related FC, the overall results suggest that our models may have been reasonably robust to demographic differences. To further help reduce overfitting, cross-validation, independent testing, and permutation-based evaluation were implemented. Leave-one-subject-out sensitivity analyses also demonstrated generally similar test-set correlation directions and magnitudes following sequential removal of individual participants. The modest sample size, however, limits conclusions regarding model stability and broader generalizability. Furthermore, as fasted-state scans always preceded fed-state scans, potential order-related effects, such as fatigue or habituation, cannot be excluded.

While antipsychotic treatment and nicotine use are known to modulate neuronal circuits involved in weight regulation and metabolism (Siafis et al., 2018; Zoli and Picciotto, 2012), their independent effects could not be directly disentangled here due to their absence in the comparison group. Within the SZ group, however, neither nicotine use nor chlorpromazine-equivalent dosing of antipsychotics differed significantly between the training and testing sets (see Supplementary Material), suggesting limited impact on within-group model performance. Notably, related work from our lab has examined resting-state FC correlates of antipsychotic-related weight-gain risk in SZ which similarly identified connectivity between the reward, sensorimotor, visual, and default mode networks (Dodd et al., 2024), highlighting overlap in FC patterns associated with adiposity- and antipsychotic-related processes while also underscoring the need for cautious interpretation. Given that cumulative antipsychotic exposure and duration of illness were not available for the present analyses, future work will be critical to disentangle the contributions of antipsychotics, nicotine use, illness chronicity, sex, head motion, and obesity to FC alterations in SZ.

In conclusion, this study suggests that obesity in SZ and in the general population may involve both shared and distinct resting-state FC patterns associated with adiposity, and that these patterns may differ across feeding states. Findings indicate that FC involving the limbic network, the triple network, and especially the reward and appetitive network may be relevant to adiposity-related processes in both individuals with and without SZ. Furthermore, FC patterns more strongly associated with adiposity in individuals with SZ compared to the general population may involve the limbic and sensorimotor networks when fasted and the visual network when fed. Altogether, these findings advance understanding of brain connectivity associated with adiposity in SZ, which may contribute to future efforts aimed at understanding and treating obesity in this vulnerable population.

Supplementary Material

1

Table 4.

Adiposity-Associated Network Connection Comparisons Across Groups

Within-Network Between-Network
SZ Fasted None Limbic-Sensorimotor, Executive Control-Salience, Limbic-Salience
Fed Visual Default-Executive Control
CG Fasted Visual Dorsal Attention-Executive Control, Salience-Visual, Dorsal
Attention-Limbic
Fed None Limbic-Sensorimotor, Executive Control-Salience
Open in a new tab

Table 4 displays the most predictive network-level connections (Summed Importance > 0.10) for adiposity across each of the four groups and models.

Abbreviations: SZ: Schizophrenia; CG: Adiposity-Matched Comparison Group.

Highlights:

  1. Brain functional connectivity (FC) was associated with adiposity in both groups.

  2. In both groups, reward, limbic, and triple network FC were related to adiposity.

  3. In SZ, limbic/sensorimotor and visual connectivity were related to adiposity.

  4. Findings suggest satiety-dependent FC patterns associated with adiposity in SZ.

Role of the Funding Source

This work was supported by the American Heart Association (24PRE1188218 [KD]), the National Institutes of Health (F31MH137968 [KD], R01MH102224 [JRT], R01DK125417 [JRT], S10OD018435 [JRT], R01DK119236 [KTL], K01DK100445 [KTL]), and the U.S. Department of Veterans Affairs (I01CX001414 [JRT], IK6CX002178 [JRT], I01CX002834 [JRT], I01CX001949 [KTL]). The funding sources had no involvement in the study design, collection, analysis and interpretation of data, or in writing the report or submitting the article for publication.

Footnotes

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Data and Code Availability

Relevant functions used to produce the presented results are available upon request.

REFERENCES

  1. Andreasen NC, 1989. The Scale for the assessment of Negative Symptoms (SANS): Conceptual and theoretical foundations. Br. J. Psychiatry 155, 49–52. [PubMed] [Google Scholar]
  2. Andrews-Hanna JR, 2012. The Brain’s Default Network and its Adaptive Role in Internal Mentation. Neuroscientist 18, 251–270. 10.1177/1073858411403316 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Annamalai A, Kosir U, Tek C, 2017. Prevalence of obesity and diabetes in patients with schizophrenia. World J Diabetes 8, 390–396. 10.4239/wjd.v8.i8.390 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Aryutova K, Stoyanov D, 2021. Pharmaco-magnetic resonance as a tool for monitoring the medication-related effects in the brain may provide potential biomarkers for psychotic disorders. Int. J. Mol. Sci. 22, 9309. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Avery JA, Carrington M, Ingeholm JE, Darcey V, Simmons WK, Hall KD, Martin A, 2025. Automatic engagement of limbic and prefrontal networks in response to food images reflects distinct information about food hedonics and inhibitory control. Commun Biol 8, 1–12. 10.1038/s42003-025-07704-w [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Avery JA, Liu AG, Ingeholm JE, Gotts SJ, Martin A, 2021. Viewing images of foods evokes taste quality-specific activity in gustatory insular cortex. Proceedings of the National Academy of Sciences 118, e2010932118. 10.1073/pnas.2010932118 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Borowitz MA, Yokum S, Duval ER, Gearhardt AN, 2020. Weight-related differences in salience, default mode, and executive function network connectivity in adolescents. Obesity (Silver Spring) 28, 1438–1446. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Brooks SJ, Cedernaes J, Schiöth HB, 2013. Increased prefrontal and parahippocampal activation with reduced dorsolateral prefrontal and insular cortex activation to food images in obesity: a meta-analysis of fMRI studies. PLoS One 8, e60393. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Burgess CR, Ramesh RN, Sugden AU, Levandowski KM, Minnig MA, Fenselau H, Lowell BB, Andermann ML, 2016. Hunger-Dependent Enhancement of Food Cue Responses in Mouse Postrhinal Cortex and Lateral Amygdala. Neuron 91, 1154–1169. 10.1016/j.neuron.2016.07.032 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Cao X, Li Q, Liu S, Li Z, Wang Y, Cheng L, Yang C, Xu Y, 2022. Enhanced Resting-State Functional Connectivity of the Nucleus Accumbens in First-Episode, Medication-Naïve Patients With Early Onset Schizophrenia. Front. Neurosci. 16. 10.3389/fnins.2022.844519 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Carnell S, Gibson C, Benson L, Ochner CN, Geliebter A, 2012. Neuroimaging and obesity: current knowledge and future directions. Obesity Reviews 13, 43–56. 10.1111/j.1467-789X.2011.00927.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Chen EY, Zeffiro TA, 2020. Hunger and BMI modulate neural responses to sweet stimuli: fMRI meta-analysis. Int J Obes 44, 1636–1652. 10.1038/s41366-020-0608-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Chen J, Perera G, Shetty H, Broadbent M, Xu Y, Stewart R, 2022. Body mass index and mortality in patients with schizophrenia spectrum disorders: a cohort study in a South London catchment area. Gen Psychiatr 35, e100819. 10.1136/gpsych-2022-100819 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Contreras-Rodríguez O, Martín-Pérez C, Vilar-López R, Verdejo-Garcia A, 2017. Ventral and dorsal striatum networks in obesity: Link to food craving and weight gain. Biol. Psychiatry 81, 789–796. [DOI] [PubMed] [Google Scholar]
  15. Cosgrove KT, Cornier M-A, McHugo M, Melanson EL, Hild A, Kronberg E, Thomas EA, Dodd K, Claus ED, Tregellas JR, Legget KT, 2025. Exercise and diet effects on delay discounting and related neurobiology in adults with overweight or obesity: A randomized trial. Obesity n/a. 10.1002/oby.24307 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Dempster P, Aitkens S, 1995. A new air displacement method for the determination of human body composition. Med Sci Sports Exerc 27, 1692–1697. [PubMed] [Google Scholar]
  17. Ding Y, Ji G, Li G, Zhang W, Hu Y, Liu L, Wang Y, Hu C, von Deneen KM, Han Y, Cui G, Wang H, Wiers CE, Manza P, Tomasi D, Volkow ND, Nie Y, Wang G-J, Zhang Y, 2020. Altered interactions among Resting-State networks in individuals with obesity. Obesity 28, 601–608. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Dodd K, Legget KT, Cornier M-A, Novick AM, McHugo M, Berman BD, Lawful BP, Tregellas JR, 2024. Relationship between functional connectivity and weight-gain risk of antipsychotics in schizophrenia. Schizophrenia Research 267, 173–181. 10.1016/j.schres.2024.03.033 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Dodd K, McHugo M, Sarabia L, Wylie KP, Legget KT, Cornier M-A, Tregellas JR, 2025. CICADA: An automated and flexible tool for comprehensive fMRI noise reduction. Imaging Neuroscience 3, IMAG.a.114. 10.1162/IMAG.a.114 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Dong D, Wang Y, Chang X, Luo C, Yao D, 2018. Dysfunction of Large-Scale Brain Networks in Schizophrenia: A Meta-analysis of Resting-State Functional Connectivity. Schizophrenia Bulletin 44, 168–181. 10.1093/schbul/sbx034 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Donofry SD, Stillman CM, Erickson KI, 2020. A review of the relationship between eating behavior, obesity and functional brain network organization. Social Cognitive and Affective Neuroscience 15, 1157–1181. 10.1093/scan/nsz085 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Dunias ZS, Van Calster B, Timmerman D, Boulesteix A-L, van Smeden M, 2024. A comparison of hyperparameter tuning procedures for clinical prediction models: A simulation study. Statistics in Medicine 43, 1119–1134. 10.1002/sim.9932 [DOI] [PubMed] [Google Scholar]
  23. Esteban O, Markiewicz CJ, Blair RW, Moodie CA, Isik AI, Erramuzpe A, Kent JD, Goncalves M, DuPre E, Snyder M, Oya H, Ghosh SS, Wright J, Durnez J, Poldrack RA, Gorgolewski KJ, 2018. FMRIPrep: a robust preprocessing pipeline for functional MRI. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Fan L, Li H, Zhuo J, Zhang Y, Wang J, Chen L, Yang Z, Chu C, Xie S, Laird AR, Fox PT, Eickhoff SB, Yu C, Jiang T, 2016. The Human Brainnetome Atlas: A New Brain Atlas Based on Connectional Architecture. Cereb Cortex 26, 3508–3526. 10.1093/cercor/bhw157 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Fields DA, Goran MI, McCrory MA, 2002. Body-composition assessment via air-displacement plethysmography in adults and children: a review123. The American Journal of Clinical Nutrition 75, 453–467. 10.1093/ajcn/75.3.453 [DOI] [PubMed] [Google Scholar]
  26. García-García I, Jurado MÁ, Garolera M, Segura B, Sala-Llonch R, Marqués-Iturria I, Pueyo R, Sender-Palacios MJ, Vernet-Vernet M, Narberhaus A, Ariza M, Junqué C, 2013. Alterations of the salience network in obesity: a resting-state fMRI study. Hum. Brain Mapp. 34, 2786–2797. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Grace AA, 2012. Dopamine system dysregulation by the hippocampus: implications for the pathophysiology and treatment of schizophrenia. Neuropharmacology 62, 1342–1348. 10.1016/j.neuropharm.2011.05.011 [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Grimm O, Kaiser S, Plichta MM, Tobler PN, 2017. Altered reward anticipation: Potential explanation for weight gain in schizophrenia? Neurosci. Biobehav. Rev. 75, 91–103. [DOI] [PubMed] [Google Scholar]
  29. Guo Y, Xia Y, Chen K, 2023. The body mass index is associated with increased temporal variability of functional connectivity in brain reward system. Front. Nutr. 10. 10.3389/fnut.2023.1210726 [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Gupta A, Osadchiy V, Mayer EA, 2020. Brain–gut–microbiome interactions in obesity and food addiction. Nat Rev Gastroenterol Hepatol 17, 655–672. 10.1038/s41575-020-0341-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Ishida R, Ishii A, Matsuo T, Minami T, Yoshikawa T, 2022. Association between eating behavior and the immediate neural activity caused by viewing food images presented in and out of awareness: A magnetoencephalography study. PLOS ONE 17, e0275959. 10.1371/journal.pone.0275959 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Kanoski SE, Grill HJ, 2017. Hippocampus contributions to food intake control: mnemonic, neuroanatomical, and endocrine mechanisms. Biol Psychiatry 81, 748–756. 10.1016/j.biopsych.2015.09.011 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Kätzel D, Wolff AR, Bygrave AM, Bannerman DM, 2020. Hippocampal Hyperactivity as a Druggable Circuit-Level Origin of Aberrant Salience in Schizophrenia. Front Pharmacol 11, 486811. 10.3389/fphar.2020.486811 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Lee H, Kwon J, Lee J, Park B, Park H, 2022. Disrupted stepwise functional brain organization in overweight individuals. Commun Biol 5, 1–9. 10.1038/s42003-021-02957-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Leenaerts N, Jongen D, Ceccarini J, Van Oudenhove L, Vrieze E, 2022. The neurobiological reward system and binge eating: A critical systematic review of neuroimaging studies. International Journal of Eating Disorders 55, 1421–1458. 10.1002/eat.23776 [DOI] [PubMed] [Google Scholar]
  36. Li G, Hu Y, Zhang W, Wang J, Ji W, Manza P, Volkow ND, Zhang Y, Wang G-J, 2023. Brain functional and structural magnetic resonance imaging of obesity and weight loss interventions. Mol Psychiatry 28, 1466–1479. 10.1038/s41380-023-02025-y [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Li S, Hu N, Zhang W, Tao B, Dai J, Gong Y, Tan Y, Cai D, Lui S, 2019. Dysconnectivity of Multiple Brain Networks in Schizophrenia: A Meta-Analysis of Resting-State Functional Connectivity. Front. Psychiatry 10. 10.3389/fpsyt.2019.00482 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Lodge DJ, Grace AA, 2011. Hippocampal dysregulation of dopamine system function and the pathophysiology of schizophrenia. Trends Pharmacol Sci 32, 507–513. 10.1016/j.tips.2011.05.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. McGrath J, Saha S, Chant D, Welham J, 2008. Schizophrenia: a concise overview of incidence, prevalence, and mortality. Epidemiol Rev 30, 67–76. 10.1093/epirev/mxn001 [DOI] [PubMed] [Google Scholar]
  40. McWhinney SR, Brosch K, Calhoun VD, Crespo-Facorro B, Crossley NA, Dannlowski U, Dickie E, Dietze LMF, Donohoe G, Du Plessis S, Ehrlich S, Emsley R, Furstova P, Glahn DC, Gonzalez-Valderrama A, Grotegerd D, Holleran L, Kircher TTJ, Knytl P, Kolenic M, Lencer R, Nenadić I, Opel N, Pfarr J-K, Rodrigue AL, Rootes-Murdy K, Ross AJ, Sim K, Škoch A, Spaniel F, Stein F, Švancer P, Tordesillas-Gutiérrez D, Undurraga J, Vázquez-Bourgon J, Voineskos A, Walton E, Weickert TW, Weickert CS, Thompson PM, van Erp TGM, Turner JA, Hajek T, 2022. Obesity and brain structure in schizophrenia - ENIGMA study in 3021 individuals. Mol Psychiatry 27, 3731–3737. 10.1038/s41380-022-01616-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Meisenzahl EM, Schmitt GJ, Scheuerecker J, Möller HJ, 2007. The role of dopamine for the pathophysiology of schizophrenia. International Review of Psychiatry 19, 337–345. 10.1080/09540260701502468 [DOI] [PubMed] [Google Scholar]
  42. Millan EZ, Ong Z, McNally GP, 2017. Chapter 5 - Paraventricular thalamus: Gateway to feeding, appetitive motivation, and drug addiction, in: Calvey T, Daniels WMU (Eds.), Progress in Brain Research, Brain Research in Addiction. Elsevier, pp. 113–137. 10.1016/bs.pbr.2017.07.006 [DOI] [PubMed] [Google Scholar]
  43. Morris JS, Dolan RJ, 2001. Involvement of Human Amygdala and Orbitofrontal Cortex in Hunger-Enhanced Memory for Food Stimuli. J. Neurosci. 21, 5304–5310. 10.1523/JNEUROSCI.21-14-05304.2001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Overall JE, 1962. The brief psychiatric rating scale. Psychol. Rep. 10, 799. [Google Scholar]
  45. Park B, Byeon K, Lee MJ, Chung C-S, Kim S-H, Morys F, Bernhardt B, Dagher A, Park H, 2020. Whole-brain functional connectivity correlates of obesity phenotypes. Human Brain Mapping 41, 4912–4924. 10.1002/hbm.25167 [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Parsons N, Steward T, Clohesy R, Almgren H, Duehlmeyer L, 2022. A systematic review of resting-state functional connectivity in obesity: Refining current neurobiological frameworks and methodological considerations moving forward. Rev. Endocr. Metab. Disord. 23, 861–879. [DOI] [PubMed] [Google Scholar]
  47. Purcell SA, Legget KT, Halliday TM, Pan Z, Creasy SA, Blankenship JM, Hild A, Tregellas JR, Melanson EL, Cornier M-A, 2022. Appetitive and metabolic responses to an exercise versus dietary intervention in adults with obesity. Transl. J. Am. Coll. Sports Med 7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Ravichandran S, Bhatt RR, Pandit B, Osadchiy V, Alaverdyan A, Vora P, Stains J, Naliboff B, Mayer EA, Gupta A, 2021. Alterations in reward network functional connectivity are associated with increased food addiction in obese individuals. Sci Rep 11, 3386. 10.1038/s41598-021-83116-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Richter A, Petrovic A, Diekhof EK, Trost S, Wolter S, Gruber O, 2015. Hyperresponsivity and impaired prefrontal control of the mesolimbic reward system in schizophrenia. J Psychiatr Res 71, 8–15. 10.1016/j.jpsychires.2015.09.005 [DOI] [PubMed] [Google Scholar]
  50. Robison AJ, Thakkar KN, Diwadkar VA, 2020. Cognition and Reward Circuits in Schizophrenia: Synergistic, Not Separate. Biological Psychiatry, Heterogeneity in Schizophrenia: Mechanisms and Symptoms 87, 204–214. 10.1016/j.biopsych.2019.09.021 [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Rolland B, Amad A, Poulet E, Bordet R, Vignaud A, Bation R, Delmaire C, Thomas P, Cottencin O, Jardri R, 2015. Resting-State Functional Connectivity of the Nucleus Accumbens in Auditory and Visual Hallucinations in Schizophrenia. Schizophrenia Bulletin 41, 291–299. 10.1093/schbul/sbu097 [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Siafis S, Tzachanis D, Samara M, Papazisis G, 2018. Antipsychotic Drugs: From Receptor-binding Profiles to Metabolic Side Effects. Curr Neuropharmacol 16, 1210–1223. 10.2174/1570159X15666170630163616 [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Simmons WK, DeVille DC, 2017. Interoceptive contributions to healthy eating and obesity. Current Opinion in Psychology, Emotion 17, 106–112. 10.1016/j.copsyc.2017.07.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Simon JJ, Biller A, Walther S, Roesch-Ely D, Stippich C, Weisbrod M, Kaiser S, 2010. Neural correlates of reward processing in schizophrenia — Relationship to apathy and depression. Schizophrenia Research 118, 154–161. 10.1016/j.schres.2009.11.007 [DOI] [PubMed] [Google Scholar]
  55. Sivakumar M, Parthasarathy S, Padmapriya T, 2024. Trade-off between training and testing ratio in machine learning for medical image processing. PeerJ Computer Science 10, e2245. 10.7717/peerj-cs.2245 [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Snoek L, Miletić S, Scholte HS, 2019. How to control for confounds in decoding analyses of neuroimaging data. NeuroImage 184, 741–760. 10.1016/j.neuroimage.2018.09.074 [DOI] [PubMed] [Google Scholar]
  57. Stip E, Lungu OV, Anselmo K, Letourneau G, Mendrek A, Stip B, Lipp O, Lalonde P, Bentaleb LA, 2012. Neural changes associated with appetite information processing in schizophrenic patients after 16 weeks of olanzapine treatment. Transl. Psychiatry 2, e128. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Subashini R, Deepa M, Padmavati R, Thara R, Mohan V, 2011. Prevalence of diabetes, obesity, and metabolic syndrome in subjects with and without schizophrenia (CURES-104). J. Postgrad. Med. 57, 272–277. [DOI] [PubMed] [Google Scholar]
  59. Takada K, Ishii A, Matsuo T, Nakamura C, Uji M, Yoshikawa T, 2018. Neural activity induced by visual food stimuli presented out of awareness: a preliminary magnetoencephalography study. Sci Rep 8, 3119. 10.1038/s41598-018-21383-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Tregellas JR, Smucny J, Harris JG, Olincy A, Maharajh K, Kronberg E, Eichman LC, Lyons E, Freedman R, 2014. Intrinsic hippocampal activity as a biomarker for cognition and symptoms in schizophrenia. Am. J. Psychiatry 171, 549–556. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Volkow ND, Wang G-J, Baler RD, 2011. Reward, dopamine and the control of food intake: implications for obesity. Trends in Cognitive Sciences 15, 37–46. 10.1016/j.tics.2010.11.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  62. Wang X, Zhang Y, Huang J, Wang Y, Niu Y, Lui SSY, Hui L, Chan RCK, 2023. Revisiting reward impairments in schizophrenia spectrum disorders: a systematic review and meta-analysis for neuroimaging findings. Psychological Medicine 53, 7189–7202. 10.1017/S0033291723000703 [DOI] [PubMed] [Google Scholar]
  63. Wang Y, Kang J, Kemmer PB, Guo Y, 2016. An Efficient and Reliable Statistical Method for Estimating Functional Connectivity in Large Scale Brain Networks Using Partial Correlation. Front Neurosci 10, 123. 10.3389/fnins.2016.00123 [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Yeo BTT, Krienen FM, Sepulcre J, Sabuncu MR, Lashkari D, Hollinshead M, Roffman JL, Smoller JW, Zöllei L, Polimeni JR, Fischl B, Liu H, Buckner RL, 2011. The organization of the human cerebral cortex estimated by intrinsic functional connectivity. Journal of Neurophysiology 106, 1125–1165. 10.1152/jn.00338.2011 [DOI] [PMC free article] [PubMed] [Google Scholar]
  65. Zeng J, Yan J, Cao H, Su Y, Song Y, Luo Y, Yang X, 2022. Neural substrates of reward anticipation and outcome in schizophrenia: a meta-analysis of fMRI findings in the monetary incentive delay task. Transl Psychiatry 12, 1–14. 10.1038/s41398-022-02201-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  66. Zoli M, Picciotto MR, 2012. Nicotinic Regulation of Energy Homeostasis. Nicotine & Tobacco Research 14, 1270–1290. 10.1093/ntr/nts159 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS2191174-supplement-1.docx (1.7MB, docx)

Data Availability Statement

Relevant functions used to produce the presented results are available upon request.

RESOURCES