BMIgap: a new tool to quantify transdiagnostic brain signatures of current and future weight

BMIgap: a new tool to quantify transdiagnostic brain signatures of current and future weight | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article BMIgap: a new tool to quantify transdiagnostic brain signatures of current and future weight Nikolaos Koutsouleris, Adyasha Tejaswi Khuntia, David Popovic, and 19 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5259910/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 20 Oct, 2025 Read the published version in Nature Mental Health → Version 1 posted You are reading this latest preprint version Abstract Understanding the neurobiological underpinnings of weight gain could reduce excess mortality and improve long-term trajectories of psychiatric disorders. We used support-vector machines and whole-brain voxel-wise grey matter volume to generate and validate a BMI predictor in healthy individuals (N = 1504) and applied it to individuals with schizophrenia (SCZ,N = 146), clinical high-risk states for psychosis (CHR,N = 213) and recent-onset depression (ROD,N = 200). We computed BMIgap (BMI predicted -BMI measured ), interrogated its brain-level overlaps with SCZ and explored whether BMIgap predicted weight gain at 1- and 2-year follow-up. SCZ (BMIgap = 1.05kg/m 2 ) and CHR individuals (BMIgap = 0.51 kg/m 2 ) showed increased and ROD individuals (BMIgap=-0.82 kg/m 2 ) decreased BMIgap. Shared brain patterns of BMI and SCZ were linked to illness duration, disease onset, and hospitalization frequency. Higher BMIgap predicted future weight gain, particularly in younger ROD individuals, and at 2-year follow-up. Therefore, we propose BMIgap as a potential brain-derived measure to stratify at-risk individuals and deliver tailored interventions for better metabolic risk control. Health sciences/Medical research/Outcomes research Health sciences/Diseases/Psychiatric disorders/Psychosis Health sciences/Diseases/Psychiatric disorders/Depression Health sciences/Diseases/Psychiatric disorders/Schizophrenia Health sciences/Biomarkers/Predictive markers BMI machine learning grey matter volume schizophrenia major depressive disorder Figures Figure 1 Figure 2 Figure 3 Introduction Globally 26% of adults are overweight (BMI of 25–30 kg/m 2 ) with an additional 13% classified as obese (BMI ≥ 30 kg/m 2 ) ( 1 ), highlighting the pandemic nature of obesity ( 2 , 3 ). Given its strong associations with metabolic diseases, including cardiovascular disease and type 2 diabetes, obesity stands as a major risk factor for somatic disorders ( 4 ). Importantly, obesity frequently parallels psychiatric disorders such as schizophrenia (SCZ), major depressive disorder (MDD), bipolar, personality and anxiety disorders ( 5 , 6 ). Psychiatric patients have a two-to-three-fold higher incidence of obesity and metabolic diseases compared to the general population, significantly contributing to their excess mortality ( 7 , 8 ). Secondary factors such as smoking, alcohol use, sedentary lifestyle, and commonly prescribed medications also significantly impact body weight and metabolic syndromes in psychiatric patients ( 9 , 10 ). Particularly atypical antipsychotics and antidepressants have been implicated in weight gain ( 11 , 12 ). Furthermore, individuals exhibiting negative symptoms—characterized by affective flattening, anhedonia and avolition—are at an increased risk for weight gain ( 13 ). In turn, weight gain and obesity can negatively impact an affected individual’s quality of life and self-esteem, thereby further exacerbating already pre-existing mental health issues and ultimately leading to a vicious circle between psychiatric symptoms, secondary disease effects and weight gain ( 14 , 15 ). In healthy populations, neuroimaging studies have identified associations between higher BMI and reduced grey matter volume (GMV) in the prefrontal, temporal, parietal and occipital cortices, the cerebellum, insula, thalamus, and amygdala ( 16 – 18 ). These associations might be related to lowered inhibitory control activity for food-related stimuli and responses resulting from a disbalance of brain activity in cognitive control regions and reward-related regions ( 17 – 19 ). These findings suggest a plausible relationship between cognitive control deficits and overeating ( 19 ) potentially resulting in higher caloric intake and obesity ( 17 ). Furthermore, in patients with SCZ, higher BMI was found to be associated with reduced GMV in the prefrontal cortex (PFC), specifically the orbitofrontal cortex (OFC), and the hippocampus ( 20 ). In patients with MDD, higher BMI was correlated with reduced GMV in the medial prefrontal regions, particularly in areas involved in impulse control and emotion regulation ( 21 ). However, a comprehensive understanding of the neurobiological underpinnings of obesity, particularly in in psychiatric disorders, is lacking as are methods to identify those at risk for weight gain ( 22 , 23 ). Therefore, personalized tools are needed to capture these interactions and enable targeted interventions—such as exercise, psychotherapy, medications, or brain stimulation—to prevent weight gain, improve treatment adherence, and reduce excess mortality ( 24 , 25 ). To better understand the neurobiological underpinnings of BMI and weight gain in early stages of affective and psychotic disorders, we developed an individualized BMI prediction model based on whole-brain GMV in healthy control (HC) individuals (HC discovery ) and validated it in a HC sample (HC validation ). We then applied this BMI predictor to individuals suffering from schizophrenia (SCZ), recent-onset depression (ROD), and clinical high-risk states for psychosis (CHR). We introduce BMIgap (Body Mass Index gap estimation) as an individualized brain-based metric, which calculates the difference between brain-estimated and measured BMI (BMI predicted -BMI measured ). Next, we correlated BMIgap with future weight changes at 1- and 2-years follow-up at the group level and used it along with additional clinical factors to predict future weight gain at the individual level. Finally, we explored the phenotypic association between BMIgap, SCZ expression (i.e. neuroanatomical similarity for SCZ), and clinical features using sparse partial least squares (SPLS). We hypothesized that (i) accurate models for individualized BMI prediction can be derived from structural brain imaging using supervised machine learning, (ii) interactions between BMI-predictive and disease-specific brain signatures result in systematic brain deviations that are captured by BMIgap, (iii) BMIgap is associated with key measures of disease severity, including age of onset, illness duration or hospitalization frequency, and (iv) BMIgap can serve as a personalized brain-based tool to assess future weight gain and identify at-risk individuals in early disease stages. Materials and methods Participants The study followed the Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis (TRIPOD) reporting guidelines ( https://www.equator-network.org/reporting-guidelines/tripod-statement/ ). We included T1-weighted MRI scans of 1504 HC individuals (HC discovery =770; HC validation =734) from four independent datasets covering 14 sites: Information eXtraction from Images (IXI; https://brain-development.org/ixi-dataset/ ), Personalized Prognostic Tools for Early Psychosis Management (PRONIA; www.pronia.eu ), Norwegian Centre for Mental Disorders Research (NORMENT; https://www.med.uio.no/norment/ ( 26 )) and the Munich Brain Imaging Database (MUC, ( 27 )) (Supplementary Methods). Moreover, three patient populations were included: SCZ individuals (N = 146) from the MUC cohort and CHR (N = 213) and ROD individuals (N = 200) from the PRONIA study (Table 1 , Table S2). All studies were approved by their local ethics committee and adhered to the ethical standards as outlined in the Declaration of Helsinki ( 28 ). Table 1 Sociodemographic differences in discovery, validation, and clinical groups. HC t/χ 2 P Clinical group F/χ 2 P Discovery Validation SCZ CHR ROD Sample (N) 770 734 0.86 b 0.35 146 213 200 776.60 b < .001 BMI, Mean (SD) 25.10 (4.03) 23.03 (2.07) 12.49 a < .001 24.02 (3.40) 23.46 (3.42) 24.01 (3.57) 13.48 c < .001 Age, Mean (SD) 41.26 (15.51) 32.24 (12.75) 12.29 a < .001 30.83 (9.97) 23.92 (5.24) 26.02 (6.37) 164.42 c < .001 Sex (Female, N (%)) 435 (56.49) 373 (50.82) 4.75 b 0.029 34.00 (23.29) 103 (48.36) 96 (48.00) 55.17 b < .001 Symptoms, Mean (SD) PANSS (total) NA NA NA NA 52.26 (29.45) 52.07 (18.22) 48.44 (14.26) 2.08 c 0.13 PANSS (positive) NA NA NA NA 11.92 (8.13) 11.31 (4.16) 8.20 (2.21) 29.09 c < .001 PANSS (negative) NA NA NA NA 14.98 (9.73) 13.09 (6.86) 12.50 (5.66) 5 c .007 PANSS (general) NA NA NA NA 25.36 (16.13) 27.66 (10.05) 27.74 (8.84) 2.19 c .113 SANS (total) NA NA NA NA 44.44 (26.52) 27.08 (24.88) 23.91 (20.95) 30.69 c < .001 BDI 3.55 (5.05) 3.27 (5.20) 0.50 a 0.616 NA 24.09 (11.75) 24.56 (12.48) 182.47 c < .001 Functioning, Mean (SD) GAF:S past month 86.30 (6.43) 87.32 (6.16) 1.54 a 0.124 NA 51.31 (11.29) 53.92 (13.17) 503.97 c < .001 GAF:D past month 85.49 (6.32) 86.51 (5.89) 1.59 a 0.113 NA 51.94 (12.) 54.14 (3.60) 385.24 c < .001 GF:S current 8.50 (0.89) 8.54 (0.71) 0.51 a 0.612 NA 6.25 (1.75) 6.39 (1.23) 157.82 c < .001 GF:R current 8.49 (0.77) 8.59 (0.67) 1.40 a 0.164 NA 5.80 (1.70) 5.97 (1.81) 150.54 c < .001 WHOQoL (total) 97.45 (33.60) 102.81 (28.91) 1.65 a 0.099 NA 71.20 (28.81) 74.92 (24.86) 39.58 c < .001 Medications Antipsychotics (N (%)) NA NA NA NA NA 60 (28.17) 27 (13.50) 12.51 b < 0.001 Antidepressants (N (%)) NA NA NA NA NA 105 (49.30) 137 (68.50) 4.23 b 0.040 CPZ-equivalent (mg) NA NA NA NA 358.9 (382.4) NA NA NA NA a χ 2 -test, b t-test, c F-test We included HC individuals aged 15–75 years with a BMI of 18.5–35 kg/m 2 without current or previous psychiatric disorders, in the discovery model by following these steps: (i) to avoid an underrepresentation of the tails of the BMI distribution and to maximize model generalizability throughout the investigated BMI range, we sampled 770 HC individuals into 33 BMI bins (0.5 BMI per bin) to approximate a uniform BMI distribution (Figure S1 a); (ii) age distribution of individuals was matched within the BMI bins to mitigate the natural correlation between age and BMI ( 29 ) (Figure S1 c). These 770 HC individuals constituted the HC discovery sample, and the remaining 734 HC individuals constituted the HC validation sample. The clinical population consisted of individuals with SCZ (N = 146), CHR (N = 213) and ROD (N = 200) (Figures S1 -S2, Supplementary Methods). MRI data acquisition and pre-processing The Participants in the IXI and PRONIA studies underwent MRI scanning at 1.5T or 3T, while NORMENT and MUC participants were examined at 1.5T MRI scanners (Supplementary Methods, Table S1 ). To facilitate between-study comparability, we applied the VBM8 pre-processing pipeline described by Opel et al ( 18 ) to produce normalized, modulated GMV tissue maps (Supplementary Methods). For computational efficiency and noise reduction, GMV images were resliced to a 3x3x3 mm 3 isotropic voxel resolution. Machine learning analysis The open-source machine learning software NeuroMiner (v1.1; https://github.com/neurominer-git/NeuroMiner_1.1 ) was used for training and application of all supervised machine learning models. To prevent information leakage between train and test data and enhance model generalizability, we implemented a repeated nested cross-validation (CV) cycle with five folds and permutations each at the inner and outer CV cycles. Thus, the process of model optimization was completely insulated from the process of model validation (Supplementary Methods). Data pre-processing included Gaussian smoothing of the neuroimaging data, partial correlation analysis to remove effects of chronological age from the GMV data, global mean offset correction to remove site effects, i.e. MR scanner effects, principal component analysis for dimensionality reduction, and voxel-wise scaling between 0 and 1 (Supplementary Methods). We used 𝜈-support vector machine regression with a linear kernel to predict BMI based on whole-brain voxel-wise GMV (71276 features) using the mean absolute error (MAE) as the optimization criterion. The statistical significance of the model was evaluated using 1000 permutations of the BMI label (α = 0.05). Predictive brain patterns were visualized at the voxel level combining the grand means of the cross-validation ratio (CVR) ( 30 ) and sign-based consistency mapping, thereby assessing whether a feature consistently predicted higher or lower BMI across CV partitions ( 32 ) (Supplementary Methods). We applied this BMI predictor to the HC validation to assess the model’s generalizability. Furthermore, the BMI predictor was applied to SCZ, ROD and CHR individuals to obtain brain-based BMI predictions for these clinical populations. BMIgap was calculated by subtracting the measured from the predicted BMI scores (BMIgap = BMI predicted –BMI measured ), where a positive BMIgap reflected a higher brain-based prediction of BMI than the measured BMI, and vice versa. To mitigate systematic bias in predicted BMI, characterized by over-estimation at lower BMI ranges and under-estimation at higher BMI ranges, BMIgap were adjusted for BMI through partial correlation analysis ( 33 ) (Supplementary Methods, Figure S3-S4). We used the corrected BMIgap for all further analysis steps. Furthermore, we independently calculated BMIgap for antipsychotic and antidepressant naïve CHR and ROD individuals to investigate whether medication influenced BMIgap in these subpopulations. Since, there were no unmedicated SCZ patients, we correlated BMIgap with their chlorpromazine equivalents. Clinical Investigation of BMIgap To understand the clinical implications of BMIgap, we analyzed its relationship with clinical variables, particularly within the SCZ group. First, to identify brain patterns distinguishing patients with SCZ from HC individuals, we classified SCZ vs. HC discovery in the MUC sample using the same pre-processing and CV settings as for the BMI-prediction model. The binarized sign-consistency maps derived from SCZ and BMI predictors were overlapped to identify brain regions commonly predictive of both phenotypes (Supplement methods). Furthermore, we extracted the decision scores from the SCZ classifier, which we refer to as SCZ expression score. A higher SCZ expression score indicates a higher likelihood to be classified as SCZ and therefore a greater neuroanatomical similarity to SCZ, while a lower SCZ expression score reflects higher likelihood of HC classification and neuroanatomical HC similarity. Next, we studied the covariation between BMIgap, SCZ likeness, and clinical variables within the overlapping brain regions of SCZ and BMI. To this end, we employed multivariate SPLS using the SPLS Toolbox by Popovic et al. ( 34 , 35 ) to investigate covariance patterns between two data domains in the SCZ sample: (i) a six feature matrix including BMIgap, SCZ expression scores, PANSS total score, age of onset, illness duration and number of hospitalizations and, (ii) a brain data matrix containing the vectorized voxels extracted using the binarized mask of overlapping BMI- and SCZ-predictive voxels. The SPLS algorithm uses singular value decomposition to generate multiple layers of distinct, multivariate associative effects between the two data matrices, called latent variables (LV) (Supplementary Methods). Investigation of BMIgap and future weight change We investigated the association between BMIgap and future weight change using two approaches: (i) correlation analysis, and (ii) ML-based prediction of weight gain. In the PRONIA cohort, where longitudinal data was available, we correlated the BMIgap to weight changes at 1- (T1) and 2-year (T2) follow-up. Weight changes were calculated as the difference between the weight at the follow-up time point and the weight at baseline (T0): 1-year (ΔW 1 = Weight T1 -Weight T0 ), and 2 years (ΔW 2 = Weight T2 -Weight T0 ). Correlation analyses were conducted for the entire cohort, as well as separately for sex and study group. In the first step, we analyzed the correlation between BMIgap and all observed weight changes. Building on previous literature, we correlated BMIgap to weight changes only within subpopulations of patients that exhibited at least + 3%, + 5% or + 7% weight gain at the respective follow-up ( 36 – 38 ). Additionally, we analyzed these correlations across different age ranges, including individuals over 15, 20, 25, 30, and 35 years, as well as within 5-year bins: 15–20, 20–25, 25–30, 30–35, and 35–40 years. Finally, we used the three weight gain thresholds (+ 3%, + 5%, and + 7%) as classification criteria to predict whether CHR and ROD individuals experienced weight gain above these thresholds at T1 and T2 or not. We used BMIgap as well as age, sex, study group (ROD, CHR), exercise (strenuous exercise or mindfulness activities such as yoga and meditation), and history of somatic comorbidities (i.e. whether the individual suffered from somatic illness) as features. Classification analyses were conducted with and without BMIgap as a feature to assess if BMIgap significantly impacts the prediction of future weight gain. Results Individualized BMI prediction The model predicted BMI in the HC discovery individuals with an MAE of 2.75 kg/m 2 (R 2 = 0.28,P < 0.001) and generalized to the HC validation with an MAE of 2.29 kg/m 2 for HC (R 2 = 0.26,P < 0.001, Fig. 1 a). Applied to the clinical subpopulations, the BMI predictor yielded an MAE of 2.85 kg/m 2 for SCZ (R 2 = 0.25,P < 0.001), an MAE of 3.07 kg/m 2 for CHR (R 2 = 0.16,P < 0.001) and an MAE of 2.73 kg/m 2 for ROD (R 2 = 0.10,P < 0.001) individuals (Table 2 ; Fig. 1 b). Table 2 Model performances of the regression analysis for the discovery model and its application to the replication and patient groups. Study group N BMIgap uncorrected (kg/m 2 ) BMIgap (kg/m 2 ) MAE (kg/m 2 ) R 2 r P HC discovery 770 -0.01 (3.4) 0 (1.78) 2.75 0.28 0.53 < 0.001 HC validation 734 1.73 (2.2) 0.23 (1.68) 2.29 0.26 0.51 < 0.001 SCZ 146 1.83 (3.0) 1.05 (1.53) 2.85 0.25 0.50 < 0.001 CHR 213 1.70 (3.26) 0.51 (1.68) 3.07 0.16 0.40 < 0.001 ROD 200 -0.03 (3.48) -0.82 (1.64) 2.73 0.10 0.32 0.001 Lower GMV in the cerebellar, prefrontal, occipital, and insular cortices, the postcentral gyrus, hippocampus, and thalamus was predictive of higher BMI. Lower GMV in the left hemisphere involving the cingulate, cerebellar, inferior occipital and temporal cortices, as well as in the right hemisphere covering parts of the precuneus, putamen and Rolandic operculum were predictive of lower BMI (Fig. 1 c, Figure S5, Table S3). BMIgap estimation across clinical groups The application of the BMI predictor yielded a mean (± SD) BMIgap of 0.23 (± 1.68) kg/m 2 for HC validation , 1.05 (± 1.53) kg/m 2 for SCZ, -0.82 (± 1.64) kg/m 2 for ROD and 0.51 (± 1.68) kg/m 2 for CHR individuals. BMIgap differed between HC discovery and HC validation individuals and clinical groups (Fig. 1 d; F [discovery vs. patients] = 33.90, P < 0.001; F [validation vs. patients] = 32.36, P < 0.001). Notably, post-hoc pairwise comparisons revealed significant differences in BMIgap between the HC discovery and HC validation groups (t = 2.62, P = 0.009) likely due to variations in BMI distributions, as indicated by a significant difference in variances (F = 3.8018, P < 0.001; Figure S1 A,C). BMIgap differed significantly between HC discovery and all clinical groups with the highest BMIgap for SCZ (t = 6.68,P < 0.001), followed by CHR (t = 3.72,P < 0.001) and the lowest for ROD (t=-5.88,P < 0.001). We did not find significant differences in BMIgap between medication naïve (N = 80 (37.56%)), antipsychotic-naïve (N = 153 (71.83%)), antidepressant-naïve (N = 108 (50.70%)) and concurrently antidepressant-antipsychotics-treated (N = 133 (62.44%)) within the CHR individuals (F = 0.6, P = 0.6244). Furthermore, we did not find significant differences between BMIgap of medication naïve (N = 59 (29.50%)), antipsychotic-naïve (N = 173 (86.50%)), antidepressant-naïve (N = 63 (31.50%)) and concurrently antidepressant-antipsychotics-treated (N = 141 (70.50%)) ROD individuals (F = 0.002, P = 0.9964). Moreover, we did not find a significant correlation between BMIgap and chlorpromazine equivalents ( r =-0.01,P = 0.86) in the SCZ sample. SCZ specific brain-signatures The SCZ classifier yielded a BAC of 72.4% (sensitivity = 72.2%, specificity = 72.6%; P < 0.001) in separating HC from SCZ individuals. Voxels predictive of SCZ were found predominantly in the inferior, middle and superior frontal gyrus, as well as in hippocampal, thalamic, insular, Rolandic operculum, postcentral, cerebellar, and basal ganglia structures. On the right hemisphere the lingual, fusiform gyrus and mid-temporal lobe were predictive of the SCZ class membership (Figure S6). The brain patterns of the BMI predictor and the SCZ classifier overlapped in the inferior, middle and superior frontal gyrus, caudate, putamen, Rolandic operculum, right precuneus and the middle temporal lobe regions (Figure S7). Clinical associations of shared SCZ and BMI signatures The SPLS analysis yielded five significant LVs, representing distinct levels of association between the neuroanatomic overlap regions of the BMI and SCZ models and the clinical disease features (Fig. 2 , Figure S8). While LV2, LV3, and LV5 captured disease-specific patterns, LV1 and LV4 extracted covariate patterns of age and sex (Supplement Results). In LV2 ( r = 0.84, P < 0.001), higher BMIgap, SCZ expression scores, age of onset, number of hospitalizations and illness duration were related to decreased GMV in the default mode network (DMN) specifically in the A, B and C subcomponents, visual, somatomotor-A, attention, salience, limbic, and control networks and increased GMV in DMN (auditory) and somatomotor B networks (Fig. 2 a). In LV3 ( r = 0.85, P < 0.001), higher BMIgap and higher SCZ expression scores were related to decreased GMV in the DMN-B and increased GMV in the DMN-D (auditory) (Fig. 2 b). In LV5 ( r = 0.58, P < 0.001), lower PANSS total score, illness duration, and age of onset and higher SCZ expression score were related to decreased GMV in DMN-C and Control-C networks, and increased GMV in the DMN-A/B, Somatomotor-B, Dorsal attention-B, and salience networks (Fig. 2 c). BMIgap and future weight change In HC individuals, BMIgap showed a positive correlation with 2-year weight gain (ΔW 2 , N = 216; r = 0.14,P < 0.05). Specifically, at T2, adults between in the age 25–40 years subgroup (N = 46; r = 0.46,P < 0.001). BMIgap of ROD individuals positively correlated with both ΔW 1 (N = 141; r = 0.18,P < 0.05) and ΔW 2 (N = 92; r = 0.3,P < 0.01) at all ages and notably for young individuals between 15–20 years (ΔW 2 :N = 15; r = 0.52,P < 0.05). Moreover, female ROD individuals showed a significant correlation between BMIgap and weight gain at T1 (ΔW 1 :N = 70; r = 0.29,P < 0.05), while male individuals did not (N = 71; r = 0.02,P = 0.86). CHR individuals between 20–40 years of age showed significant correlations between BMIgap and ΔW 1 in the + 5% weight increase subgroup (N = 58; r = 0.20,P < 0.05) as well as ΔW 2 in the + 3% (N = 39; r = 0.28,P < 0.05) and + 5% (N = 32; r = 0.36,P < 0.05) subgroup (Fig. 3 A, Figure S10). Weight gain predictor The multivariate weight gain prediction model indicated a weight gain of + 7% at T2 with a BAC of 59.2% (sensitivity = 64.9%, specificity = 53.5%, P < 0.05), while the + 3% (BAC = 52.2%; sensitivity = 52.5%, specificity = 51.9%, P = 0.21) and + 5% weight gain prediction models did not perform above chance level (BAC = 45.2%, sensitivity = 62.5%, specificity = 35.2%) (Table S6). Key predictive features for the 7% weight gain predictor included age, clinical group (ROD, CHR), exercise, somatic history, BMI gap, and tobacco use. Specifically, amongst CHR individuals, a higher BMIgap, in combination with a history of somatic conditions and reduced mindfulness-based exercises was predictive of weight gain. For ROD individuals, the BMIgap was negatively associated with 7% weight gain, particularly in individuals with history of lower somatic comorbidities, reduced tobacco use, and higher age (Fig. 3 C). Notably, the predictive performance of the + 7% weight reduced after excluding BMIgap as a feature (BAC = 56.1%; sensitivity = 48.6%, specificity = 63.6%; P < 0.05) and these two models differed significantly (P < 0.05) (Fig. 3 C). Discussion The aim of our study was to introduce BMIgap as a metric to evaluate BMI-related brain signatures, interrogate its overlaps with SCZ brain patterns and explore its implications for future weight gain. The BMI predictor relied on lower GMV in frontal and temporal regions and higher GMV in precuneus, putamen and Rolandic operculum to predict BMI. Individuals in the psychosis spectrum (SCZ and CHR) showed an increased BMIgap while individuals suffering from affective diseases (ROD) showed a decreased BMIgap. Moreover, we found that separability of SCZ from HC individuals partly overlapped with BMI-related structural brain variation pertaining to the inhibitory control and reward systems. SPLS analysis revealed that a prefronto-temporal brain pattern predicting both disease and BMI phenotypes was associated with longer illness duration, later disease onset, and higher number of hospitalizations. The potential clinical utility of increased BMIgap, i.e., having a higher brain-estimated BMI than currently measured, was elucidated by our post-hoc analyses. On a group level, higher BMIgap was correlated with future weight gain, an effect particularly pronounced in longer-term trajectories of younger individuals, who were suffering from depressive disorders. On an individual level, BMIgap was identified as a predictive feature for future weight gain in combination with age, clinical group (ROD and CHR), exercise, history of somatic comorbidities, and tobacco usage. The co-occurrence of higher BMI and lower GMV in the reward and salience systems in the BMI predictor model may represent a neurobiological mediator of eating behaviors ( 16 – 19 ). Furthermore, BMI-predictive GMV reductions in brain areas related to taste, reward, and inhibitory control may contribute to increased susceptibility to hypercaloric eating ( 17 , 19 ). In patient groups, the SCZ signature and the BMI signatures overlapped within inhibitory, reward and cognitive control regions, thereby pointing to interacting pathways between obesity and SCZ consistent with previous research ( 20 , 39 ). Our findings substantiate previous reports that prefrontal deficits might lead to reduced cognitive control and therefore may amplify the risk of addictive behaviors such as overeating which could significantly contribute to an increased BMI in psychiatric patients ( 17 , 18 , 39 ). Positive BMIgap in SCZ and CHR groups indicate that these patients have brain alterations typically found in healthy individuals with higher BMI. This may imply that underlying pathophysiological processes—such as inflammation, insulin resistance, and disruptions in gut microbiota—may influence brain structure similarly to the effects seen in higher BMI conditions and ultimately predispose them to future weight gain ( 40 , 41 ). These interaction effects may be more pronounced in SCZ compared to CHR individuals given their more advanced disease stage, resulting in higher BMIgap and a higher propensity towards unfavorable metabolic outcomes. Conversely, a negative BMIgap in the ROD group resembles brain signatures found in individuals with a lower BMI. This is supported by previous findings showing overlapping brain patterns between depressed and underweight individuals as well as patients with comorbid depression and anorexia nervosa ( 42 ). Moreover, occurrence of metabolic syndromes in psychiatric patients is often linked to psychiatric medication, namely antipsychotics and antidepressants ( 11 , 43 , 44 ). Antipsychotics impact the mesolimbic dopaminergic system and the ventromedial nucleus, thereby altering behavioral responses to environmental stimuli and regulating both food intake and bodyweight ( 43 ). While antidepressants impact metabolic risk by increasing appetite and suppressing satiety, primarily through histaminergic and serotonin receptor antagonism ( 11 , 12 ). However, we did not find significant differences when comparing the BMIgap of fully naïve, partly naïve and concurrently antidepressant-antipsychotics-treated CHR or ROD. Moreover, we did not find any significant correlations between BMIgap and antipsychotic dosage in SCZ patients. These findings indicate that the BMIgap captures a neurobiological BMI signature, which may represent a more disease-specific or individual predisposition toward future weight changes. This finding aligns with previous research indicating that patients with psychiatric disorders, are at an increased risk of developing metabolic syndrome and obesity, independent of medication use ( 45 , 46 ), contrasting other literature that primarily attributes weight gain to the effects of psychiatric medication ( 11 , 12 ). The phenotypic association analysis linked control and reward brain networks to SCZ diagnosis, BMIgap and clinical variables in unique ways. The multivariate signatures of concurrently high BMIgap and SCZ expression scores (LV2, LV3) were associated with both a decrease in GMV within the limbic network and increased GMV within the DMN (auditory) network. In obesity research, these networks have been involved in reward processing ( 47 ), food motivation ( 48 ), executive and affective control ( 45 ), while in SCZ research, they have been particularly implicated in the impaired processing of negative emotions ( 49 ). Furthermore, later disease onset, higher hospitalization frequency, and longer illness duration, were associated with both higher BMIgap and SCZ-expression scores, thereby implying a potential association between the severity of SCZ and the presence of obesity risk traits. Moreover, we identified a pattern independent of BMIgap (LV5) suggesting that there are distinct effects of SCZ and BMI on the brain, indicating that those with early onset, shorter illness duration, and milder symptoms are less likely to show high SCZ-diagnostic separability, thus, highlighting the complexity of SCZ subtypes and the variability in its manifestation and progression. Moreover, our assertion that the BMIgap is associated with future weight changes was supported by the positive correlation between BMIgap and future weight change observed across all groups, with the strength of this correlation increasing from 1 to 2-year follow-up. Notably, these correlations were most pronounced among young individuals (15–20 years) in the ROD group and young adults (25–40 years) in the CHR group, indicating that individuals in the early stages of psychiatric illness may exhibit distinct brain signatures that predispose them to future weight gain. Furthermore, the 7% weight gain predictor, highlighted that ROD young adults who have higher exercise rates, use less tobacco, have less somatic comorbidities, and a lower BMIgap are less likely to experience weight gain. Conversely, younger CHR individuals with reduced exercise rates, more history of somatic comorbidities, higher tobacco use, and a higher BMIgap were at an increased risk for future weight gain. Notably, the exclusion of the BMIgap feature from the 7% weight gain predictor resulted in a significant reduction of predictive accuracy of the model in correctly predicting future weight gain. Therefore, the BMIgap may be helpful in identifying patients at risk of weight gain, facilitating the early implementation of tailored interventions such as weight-neutral medications, individualized dietary and physical activity regimens, to stabilize weight ( 24 , 25 ). Limitations Among the limitations of our study is the lack of individuals with very high (> 35) or low (< 18) BMI, which should be further addressed by future studies focusing on the understudied subpopulations of underweight and highly obese individuals. Moreover, the study lacked metabolic markers necessary to further interrogate the shared pathophysiological substrates affecting brain structure and clinical covariations between obesity and psychiatric disorders. However, BMI, being widely used in similar studies, enabled us to make preliminary and direct comparisons with prior findings ( 16 – 18 , 39 , 50 ). Conclusion In conclusion, our study identified BMIgap as a crucial metric for exploring the relationship between BMI and brain structure in psychiatric populations, particularly within the depression and schizophrenia spectrum. We found that elevated BMIgap was associated with significant neurobiological alterations in reward and inhibitory control systems, indicating a complex interplay between obesity and schizophrenia. Our findings further suggest that BMIgap could potentially be a predictive indicator of future weight gain, especially among younger individuals with a higher disease burden. Therefore, BMIgap could serve as a template of using machine learning and brain imaging to enhance early identification of patients at risk for metabolic complications. Integrating BMIgap, or future, more sophisticated tools, into clinical assessments may improve strategies for preventing weight gain in psychiatric patients. Declarations Acknowledgements Author contributions: AK, AW and NK had access to all the study data. NK and ESc conceived and designed the study. AK, DP, AW, and NK were responsible for the analytic framework and data interpretation. AK was responsible for data organization. MLP provided the NORMENT data. NK, JK, RKRS, JH, RL, AB, PB, RU, SB, SW, RL, PF, LW, OA and EM were responsible for the data collection. AK, AW, DP and ESa were responsible for the analysis set-up. AK wrote the first manuscript draft critically revised by DP, AW, and NK. ESc, RU, RL, LW, OAA, MB and ES critically revised the manuscript for important intellectual content. All authors read and approved the final manuscript. We would like to thank Susanne Miedl for helping with the administrative work. Funding: This study was supported by the German Federal Ministry of Education and Research (BMBF, 01ZX1904E and 01ZX2204A) as a part of COMorbidity Modeling via Integrative Transfer machine learning in MENTal illness project. PRONIA is a collaboration project funded by the European Union under the 7th Framework Programme under grant agreement no. 602152. NK is supported through grants from NIH (U01MH124639-01; ProNET), the Wellcome Trust, the German Innovation Fund (CARE project), the German Federal Ministry of Education and Research (COMMITMENT and BEST projects), as well as ERA PerMed (IMPLEMENT project). AK is funded through the COMMITMENT project. DP was supported by the Else-Kröner-Fresenius-Foundation through the Clinician Scientist Program ‘EKFS-Translational Psychiatry’. ESc is supported through grants by the German Federal Ministry of Education and Research (COMMITMENT (grant 01ZX2204A), BEST (grant 01EK2101B), and IMPLEMENT (grant 01KU1905A) projects). The Norwegian study group is funded by the Research Council of Norway and the KG Jebsen stiftelsen. Role of funders: The funders were not involved in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. Competing Interests NK owns issued patent US20160192889A1 (‘Adaptive pattern recognition for psychosis risk modeling’). As a member of Spring Health's Scientific Advisory Board, NK has advised that company on the development of tools to predict treatment outcomes for depression and psychosis. He has no equity and has received no financial compensation from this company. AB reports speaker fees from Otsuka, Lundbeck, Angelini and Rovi outside of the submitted work. JH reports personal fees from Orion ltd, personal fees from Lundbeck, personal fees from Otsuka and others from Takeda during the conduct of the study. RU reports speaker fees from Sunovion, Otsuka and Vitaris outside the submitted work as well as unpaid officership with the British Association for Pharmacology - Honorary General Secretary 2021-2024. She serves as Deputy Editor for The British Journal of Psychiatry. PF reports he has received research support/honoraria for lectures or advisory activities from Boehringer-Ingelheim, Janssen, Lundbeck, Otsuka, Recordati and Richter outside the submitted work. OAA is a consultant to Cortechs.ai and has received speaker’s honorarium from Otsuka, Lundbeck, Janssen and Sunovion. RL reports honoraria for lectures or advisory activities from Boehringer-Ingelheim, Janssen, Otsuka, and ROVI outside the submitted work. The remaining authors have nothing to disclose. References Organization WH (2005): Surveillance of Chronic Disease Risk Factors : Country Level Data and Comparable Estimates . World Health Organization. Retrieved May 19, 2024, from https://iris.who.int/handle/10665/43190 Heymsfield SB, Wadden TA (2017): Mechanisms, Pathophysiology, and Management of Obesity. N Engl J Med 376: 254–266. Weiss F, Barbuti M, Carignani G, Calderone A, Santini F, Maremmani I, Perugi G (2020): Psychiatric Aspects of Obesity: A Narrative Review of Pathophysiology and Psychopathology. J Clin Med 9: 2344. Khafagy R, Dash S (2021): Obesity and Cardiovascular Disease: The Emerging Role of Inflammation. Front Cardiovasc Med 8: 768119. Rajan T, Menon V (2017): Psychiatric disorders and obesity: A review of association studies. J Postgrad Med 63: 182–190. Afzal M, Siddiqi N, Ahmad B, Afsheen N, Aslam F, Ali A, et al. (2021): Prevalence of Overweight and Obesity in People With Severe Mental Illness: Systematic Review and Meta-Analysis. Front Endocrinol 12. Retrieved March 3, 2023, from https://www.frontiersin.org/articles/10.3389/fendo.2021.769309 Abdullah HM, Shahul HA, Hwang MY, Ferrando S (2020): Comorbidity in Schizophrenia: Conceptual Issues and Clinical Management. Focus J Life Long Learn Psychiatry 18: 386. De Micheli A, Provenzani U, Krakowski K, Oliver D, Damiani S, Brondino N, et al. (2024): Physical Health and Transition to Psychosis in People at Clinical High Risk [no. 3]. Biomedicines 12: 523. Kar N, Barreto S (2024): Influence of Lifestyle Factors on Metabolic Syndrome in Psychiatric Patients Attending a Community Mental Health Setting: A Cross-sectional Study. Indian J Psychol Med 46: 313–322. Sadeghirad B, Duhaney T, Motaghipisheh S, Campbell NRC, Johnston BC (2016): Influence of unhealthy food and beverage marketing on children’s dietary intake and preference: a systematic review and meta-analysis of randomized trials. Obes Rev Off J Int Assoc Study Obes 17: 945–959. H G, B G, S E-H, D C-L, O L, Jd R, et al. (2020): Antidepressant Medications and Weight Change: A Narrative Review. Obes Silver Spring Md 28. https://doi.org/10.1002/oby.22969 A S, L M (2010): Antidepressants and body weight: a comprehensive review and meta-analysis. J Clin Psychiatry 71. https://doi.org/10.4088/JCP.09r05346blu Hryhorczuk C, Sharma S, Fulton SE (2013): Metabolic disturbances connecting obesity and depression. Front Neurosci 7: 177. Sarwer DB, Polonsky HM (2016): The Psychosocial Burden of Obesity. Endocrinol Metab Clin North Am 45: 677–688. Lee K, Akinola A, Abraham S (n.d.): Antipsychotic-induced weight gain: exploring the role of psychiatrists in managing patients’ physical health – challenges, current options and direction for future care. BJPsych Bull 48: 24–29. Li L, Yu H, Zhong M, Liu S, Wei W, Meng Y, et al. (2022): Gray matter volume alterations in subjects with overweight and obesity: Evidence from a voxel-based meta-analysis. Front Psychiatry 13. Retrieved March 3, 2023, from https://www.frontiersin.org/articles/10.3389/fpsyt.2022.955741 Fernández-Andújar M, Morales-García E, García-Casares N (2021): Obesity and Gray Matter Volume Assessed by Neuroimaging: A Systematic Review. Brain Sci 11: 999. Opel N, Redlich R, Kaehler C, Grotegerd D, Dohm K, Heindel W, et al. (2017): Prefrontal gray matter volume mediates genetic risks for obesity. Mol Psychiatry 22: 703–710. Stice E, Figlewicz DP, Gosnell BA, Levine AS, Pratt WE (2013): The contribution of brain reward circuits to the obesity epidemic. Neurosci Biobehav Rev 37: 10.1016/j.neubiorev.2012.12.001. Tsai S-Y, Sajatovic M, Hsu J-L, Chung K-H, Chen P-H, Huang Y-J (2020): Body mass index, residual psychotic symptoms, and inflammation associated with brain volume reduction in older patients with schizophrenia. Int J Geriatr Psychiatry 35: 728–736. Opel N, Thalamuthu A, Milaneschi Y, Grotegerd D, Flint C, Leenings R, et al. (2021): Brain structural abnormalities in obesity: relation to age, genetic risk, and common psychiatric disorders. Mol Psychiatry 26: 4839–4852. Hinchliffe N, Capehorn MS, Bewick M, Feenie J (2022): The Potential Role of Digital Health in Obesity Care. Adv Ther 39: 4397–4412. Bays HE, Fitch A, Cuda S, Gonsahn-Bollie S, Rickey E, Hablutzel J, et al. (2023): Artificial intelligence and obesity management: An Obesity Medicine Association (OMA) Clinical Practice Statement (CPS) 2023. Obes Pillars 6: 100065. Tian YE, Di Biase MA, Mosley PE, Lupton MK, Xia Y, Fripp J, et al. (2023): Evaluation of Brain-Body Health in Individuals With Common Neuropsychiatric Disorders. JAMA Psychiatry 80: 567–576. Alhindi YA, Khalifa N, Al-Khyatt W, Idris I (2023): The use of non-invasive brain stimulation techniques to reduce body weight and food cravings: A systematic review and meta-analysis. Clin Obes 13: e12611. Wolfers T, Doan NT, Kaufmann T, Alnæs D, Moberget T, Agartz I, et al. (2018): Mapping the Heterogeneous Phenotype of Schizophrenia and Bipolar Disorder Using Normative Models. JAMA Psychiatry 75: 1146–1155. Koutsouleris N, Meisenzahl EM, Borgwardt S, Riecher-Rössler A, Frodl T, Kambeitz J, et al. (2015): Individualized differential diagnosis of schizophrenia and mood disorders using neuroanatomical biomarkers. Brain J Neurol 138: 2059–2073. Goodyear MDE, Krleza-Jeric K, Lemmens T (2007): The Declaration of Helsinki. BMJ 335: 624–625. Stevens J, Cai J, Pamuk ER, Williamson DF, Thun MJ, Wood JL (1998): The Effect of Age on the Association between Body-Mass Index and Mortality. N Engl J Med 338: 1–7. Koutsouleris N, Kambeitz-Ilankovic L, Ruhrmann S, Rosen M, Ruef A, Dwyer DB, et al. (2018): Prediction Models of Functional Outcomes for Individuals in the Clinical High-Risk State for Psychosis or With Recent-Onset Depression: A Multimodal, Multisite Machine Learning Analysis. JAMA Psychiatry 75: 1156–1172. Koutsouleris N, Dwyer DB, Degenhardt F, Maj C, Urquijo-Castro MF, Sanfelici R, et al. (2021): Multimodal Machine Learning Workflows for Prediction of Psychosis in Patients With Clinical High-Risk Syndromes and Recent-Onset Depression. JAMA Psychiatry 78: 195–209. Gómez-Verdejo V, Parrado-Hernández E, Tohka J, Alzheimer’s Disease Neuroimaging Initiative (2019): Sign-Consistency Based Variable Importance for Machine Learning in Brain Imaging. Neuroinformatics 17: 593–609. Zhang B, Zhang S, Feng J, Zhang S (2023): Age-level bias correction in brain age prediction. NeuroImage Clin 37: 103319. Monteiro JM, Rao A, Shawe-Taylor J, Mourão-Miranda J (2016): A multiple hold-out framework for Sparse Partial Least Squares. J Neurosci Methods 271: 182–194. Popovic D, Ruef A, Dwyer DB, Antonucci LA, Eder J, Sanfelici R, et al. (2020): Traces of Trauma: A Multivariate Pattern Analysis of Childhood Trauma, Brain Structure, and Clinical Phenotypes. Biol Psychiatry 88: 829–842. De Hert M, Dekker JM, Wood D, Kahl KG, Holt RIG, Möller H-J (2009): Cardiovascular disease and diabetes in people with severe mental illness position statement from the European Psychiatric Association (EPA), supported by the European Association for the Study of Diabetes (EASD) and the European Society of Cardiology (ESC). Eur Psychiatry J Assoc Eur Psychiatr 24: 412–424. Svärd A, Lahti J, Mänty M, Roos E, Rahkonen O, Lahelma E, Lallukka T (2020): Weight change among normal weight, overweight and obese employees and subsequent diagnosis-specific sickness absence: A register-linked follow-up study. Scand J Public Health 48: 155–163. Eder J, Glocker C, Barton B, Sarisik E, Popovic D, Lämmermann J, et al. (2024): Who is at risk for weight gain after weight-gain associated treatment with antipsychotics, antidepressants, and mood stabilizers: A machine learning approach. Acta Psychiatr Scand . https://doi.org/10.1111/acps.13684 McWhinney SR, Brosch K, Calhoun VD, Crespo-Facorro B, Crossley NA, Dannlowski U, et al. (2022): Obesity and brain structure in schizophrenia - ENIGMA study in 3021 individuals. Mol Psychiatry 27: 3731–3737. Baothman OA, Zamzami MA, Taher I, Abubaker J, Abu-Farha M (2016): The role of Gut Microbiota in the development of obesity and Diabetes. Lipids Health Dis 15: 108. Wu H, Liu Y, Wang J, Chen S, Xie L, Wu X (2023): Schizophrenia and obesity: May the gut microbiota serve as a link for the pathogenesis? iMeta 2: e99. Bohon C, Welch H (2021): Quadratic relations of BMI with depression and brain volume in children: Analysis of data from the ABCD Study. J Psychiatr Res 136: 421–427. Panariello F, De Luca V, de Bartolomeis A (2011): Weight Gain, Schizophrenia and Antipsychotics: New Findings from Animal Model and Pharmacogenomic Studies. Schizophr Res Treat 2011: 459284. Fava M (2000): Weight gain and antidepressants. J Clin Psychiatry 61 Suppl 11: 37–41. Ferns G (2018): Cause, consequence or coincidence: The relationship between psychiatric disease and metabolic syndrome. Transl Metab Syndr Res 1: 23–38. Penninx BWJH, Lange SMM (2018): Metabolic syndrome in psychiatric patients: overview, mechanisms, and implications. Dialogues Clin Neurosci 20: 63–73. Avery JA, Powell JN, Breslin FJ, Lepping RJ, Martin LE, Patrician TM, et al. (2017): Obesity is associated with altered mid-insula functional connectivity to limbic regions underlying appetitive responses to foods. J Psychopharmacol Oxf Engl 31: 1475–1484. Dugré JR, Bitar N, Dumais A, Potvin S (2019): Limbic Hyperactivity in Response to Emotionally Neutral Stimuli in Schizophrenia: A Neuroimaging Meta-Analysis of the Hypervigilant Mind. Am J Psychiatry 176: 1021–1029. Tamminga CA, Thaker GK, Buchanan R, Kirkpatrick B, Alphs LD, Chase TN, Carpenter WT (1992): Limbic System Abnormalities Identified in Schizophrenia Using Positron Emission Tomography With Fluorodeoxyglucose and Neocortical Alterations With Deficit Syndrome. Arch Gen Psychiatry 49: 522–530. Mj H, Ak T, J B, M B, B W (2019): Grey matter alterations in obesity: A meta-analysis of whole-brain studies. Obes Rev Off J Int Assoc Study Obes 20. https://doi.org/10.1111/obr.12799 Thomas Yeo BT, Krienen FM, Sepulcre J, Sabuncu MR, Lashkari D, Hollinshead M, et al. (2011): The organization of the human cerebral cortex estimated by intrinsic functional connectivity. J Neurophysiol 106: 1125–1165. Baker JT, Holmes AJ, Masters GA, Yeo BTT, Krienen F, Buckner RL, Öngür D (2014): Disruption of Cortical Association Networks in Schizophrenia and Psychotic Bipolar Disorder. JAMA Psychiatry 71: 109. Additional Declarations Yes there is potential Competing Interest. NK owns issued patent US20160192889A1 (‘Adaptive pattern recognition for psychosis risk modeling’). As a member of Spring Health's Scientific Advisory Board, NK has advised that company on the development of tools to predict treatment outcomes for depression and psychosis. He has no equity and has received no financial compensation from this company. AB reports speaker fees from Otsuka, Lundbeck, Angelini and Rovi outside of the submitted work. JH reports personal fees from Orion ltd, personal fees from Lundbeck, personal fees from Otsuka and others from Takeda during the conduct of the study. RU reports speaker fees from Sunovion, Otsuka and Vitaris outside the submitted work as well as unpaid officership with the British Association for Pharmacology - Honorary General Secretary 2021-2024. She serves as Deputy Editor for The British Journal of Psychiatry. PF reports he has received research support/honoraria for lectures or advisory activities from Boehringer-Ingelheim, Janssen, Lundbeck, Otsuka, Recordati and Richter outside the submitted work. OAA is a consultant to Cortechs.ai and has received speaker’s honorarium from Otsuka, Lundbeck, Janssen and Sunovion. RL reports honoraria for lectures or advisory activities from Boehringer-Ingelheim, Janssen, Otsuka, and ROVI outside the submitted work. The remaining authors have nothing to disclose. Supplementary Files GMVBMIsupplementalmaterialsubmition.docx Supplementary Material Cite Share Download PDF Status: Published Journal Publication published 20 Oct, 2025 Read the published version in Nature Mental Health → Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-5259910","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":388715328,"identity":"87954ed2-204f-402e-969f-1b62a8e18360","order_by":0,"name":"Nikolaos Koutsouleris","email":"data:image/png;base64,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","orcid":"https://orcid.org/0000-0001-6825-6262","institution":"Ludwig-Maximilians-University","correspondingAuthor":true,"prefix":"","firstName":"Nikolaos","middleName":"","lastName":"Koutsouleris","suffix":""},{"id":388715329,"identity":"1a6f7ee1-77fd-4a0e-8cc7-7ae05c894891","order_by":1,"name":"Adyasha Tejaswi Khuntia","email":"","orcid":"","institution":"Ludwig-Maximilian-University","correspondingAuthor":false,"prefix":"","firstName":"Adyasha","middleName":"Tejaswi","lastName":"Khuntia","suffix":""},{"id":388715330,"identity":"27becc62-3a27-4725-a671-6e88e5810ef0","order_by":2,"name":"David Popovic","email":"","orcid":"https://orcid.org/0000-0002-2367-9437","institution":"Max-Planck-Institute of Psychiatry","correspondingAuthor":false,"prefix":"","firstName":"David","middleName":"","lastName":"Popovic","suffix":""},{"id":388715331,"identity":"1bd251e4-e419-4dac-b562-c581bb53ac96","order_by":3,"name":"Elif Sarisik","email":"","orcid":"","institution":"Ludwig-Maximilians-University","correspondingAuthor":false,"prefix":"","firstName":"Elif","middleName":"","lastName":"Sarisik","suffix":""},{"id":388715332,"identity":"482bff41-2e1d-4f08-908a-8a936f149c3f","order_by":4,"name":"Madalina O. Buciuman","email":"","orcid":"","institution":"Ludwig-Maximilians-University","correspondingAuthor":false,"prefix":"","firstName":"Madalina","middleName":"O.","lastName":"Buciuman","suffix":""},{"id":388715333,"identity":"e6073c13-272a-4196-97d3-36252e164a01","order_by":5,"name":"Mads L. Pedersen","email":"","orcid":"","institution":"University of Oslo","correspondingAuthor":false,"prefix":"","firstName":"Mads","middleName":"L.","lastName":"Pedersen","suffix":""},{"id":388715334,"identity":"00e20fcd-700f-482a-9329-edaf7c3e6c6d","order_by":6,"name":"Lars T. Westlye","email":"","orcid":"https://orcid.org/0000-0001-8644-956X","institution":"University of Oslo","correspondingAuthor":false,"prefix":"","firstName":"Lars","middleName":"T.","lastName":"Westlye","suffix":""},{"id":388715335,"identity":"fdcf099a-7790-47f2-a63e-c33a2987860b","order_by":7,"name":"Ole Andreassen","email":"","orcid":"https://orcid.org/0000-0002-4461-3568","institution":"Oslo University Hospital \u0026 Institute of Clinical Medicine, University of Oslo","correspondingAuthor":false,"prefix":"","firstName":"Ole","middleName":"","lastName":"Andreassen","suffix":""},{"id":388715336,"identity":"8966eef5-ebc4-4037-998b-bbef7cf53c73","order_by":8,"name":"Andreas Meyer-Lindenberg","email":"","orcid":"","institution":"Central Institute of Mental Health, Mannheim","correspondingAuthor":false,"prefix":"","firstName":"Andreas","middleName":"","lastName":"Meyer-Lindenberg","suffix":""},{"id":388715337,"identity":"d2d1f8f1-22b0-4a80-9b6b-c96ef150f7e3","order_by":9,"name":"Joseph Kambeitz","email":"","orcid":"https://orcid.org/0000-0002-8988-3959","institution":"Faculty of Medicine and University Hospital University of Cologne Cologne","correspondingAuthor":false,"prefix":"","firstName":"Joseph","middleName":"","lastName":"Kambeitz","suffix":""},{"id":388715338,"identity":"1269f472-9db2-47b4-89e1-319933d7dbcb","order_by":10,"name":"Raimo Salokangas","email":"","orcid":"","institution":"Department of Psychiatry, University of Turku","correspondingAuthor":false,"prefix":"","firstName":"Raimo","middleName":"","lastName":"Salokangas","suffix":""},{"id":388715339,"identity":"11c3bda7-aa38-46a3-87ed-196dceb8ab32","order_by":11,"name":"Jarmo Hietala","email":"","orcid":"","institution":"Department of Psychiatry, University of Turku and Turku University Hospital, Finland","correspondingAuthor":false,"prefix":"","firstName":"Jarmo","middleName":"","lastName":"Hietala","suffix":""},{"id":388715340,"identity":"267e08d2-232b-423d-a628-419773dd471b","order_by":12,"name":"Alessandro Bertolino","email":"","orcid":"","institution":"Department of Translational Biomedicine and Neuroscience, University of Bari \"Aldo Moro\", Bari, Italy","correspondingAuthor":false,"prefix":"","firstName":"Alessandro","middleName":"","lastName":"Bertolino","suffix":""},{"id":388715341,"identity":"6f4ee47a-a5ce-4cda-b47e-3dd4afe5e06d","order_by":13,"name":"Stefan Borgwardt","email":"","orcid":"https://orcid.org/0000-0002-5792-3987","institution":"University of Lübeck","correspondingAuthor":false,"prefix":"","firstName":"Stefan","middleName":"","lastName":"Borgwardt","suffix":""},{"id":388715342,"identity":"591cc422-0df8-474f-bfc3-d8346e5ee83a","order_by":14,"name":"Paolo Brambilla","email":"","orcid":"","institution":"University of Milan","correspondingAuthor":false,"prefix":"","firstName":"Paolo","middleName":"","lastName":"Brambilla","suffix":""},{"id":388715343,"identity":"e192097e-a881-4557-a132-1791090cb6d9","order_by":15,"name":"Rachel Upthegrove","email":"","orcid":"https://orcid.org/0000-0001-8204-5103","institution":"University of Birmingham","correspondingAuthor":false,"prefix":"","firstName":"Rachel","middleName":"","lastName":"Upthegrove","suffix":""},{"id":388715344,"identity":"6241ff4d-2fdd-4b39-9915-d734a53dca85","order_by":16,"name":"Stephen Wood","email":"","orcid":"","institution":"University of Melbourne","correspondingAuthor":false,"prefix":"","firstName":"Stephen","middleName":"","lastName":"Wood","suffix":""},{"id":388715345,"identity":"d83e3dbe-342d-45fe-8a5f-b96bb9aad402","order_by":17,"name":"Rebekka Lencer","email":"","orcid":"","institution":"Department of Psychiatry and Psychotherapy and Center for Brain, Behaviour and Metabolism, University of Lübeck","correspondingAuthor":false,"prefix":"","firstName":"Rebekka","middleName":"","lastName":"Lencer","suffix":""},{"id":388715346,"identity":"a3fda4ac-e299-4671-8b00-13da029841a2","order_by":18,"name":"Eva Meisenzahl","email":"","orcid":"","institution":"Kliniken der Heinrich-Heine-Universität","correspondingAuthor":false,"prefix":"","firstName":"Eva","middleName":"","lastName":"Meisenzahl","suffix":""},{"id":388715347,"identity":"abc91490-116d-486b-9cc7-962958469600","order_by":19,"name":"Peter Falkai","email":"","orcid":"","institution":"Ludwig Maximilians University Munich","correspondingAuthor":false,"prefix":"","firstName":"Peter","middleName":"","lastName":"Falkai","suffix":""},{"id":388715348,"identity":"7942e9b2-cb48-4258-9e9c-e99c553e5b03","order_by":20,"name":"Emanuel Schwarz","email":"","orcid":"https://orcid.org/0000-0002-0261-5063","institution":"Central Institute of Mental Health","correspondingAuthor":false,"prefix":"","firstName":"Emanuel","middleName":"","lastName":"Schwarz","suffix":""},{"id":388715349,"identity":"6dfbfff2-a043-4426-8d76-3c560b2e3c64","order_by":21,"name":"Ariane Wiegand","email":"","orcid":"","institution":"Ludwig-Maximilian-University","correspondingAuthor":false,"prefix":"","firstName":"Ariane","middleName":"","lastName":"Wiegand","suffix":""}],"badges":[],"createdAt":"2024-10-14 09:30:14","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5259910/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5259910/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s44220-025-00522-3","type":"published","date":"2025-10-20T04:00:00+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":71138523,"identity":"85acccd9-81e1-47b2-9a10-3ddd79272675","added_by":"auto","created_at":"2024-12-11 13:30:47","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":329557,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eVisualization and performance of the BMI predictor\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-5259910/v1/a2328a92a09e67b356d95e9d.png"},{"id":71138525,"identity":"97a4506b-a5ce-428f-8069-6678fac35d3f","added_by":"auto","created_at":"2024-12-11 13:30:47","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":266410,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSPLS Analysis of clinical domains and shared SCZ-BMI brain regions \u003c/strong\u003e(A) LV2, (B) LV3 and (C) a BMIgap-independent signature of LV5.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-5259910/v1/aa550b63b2cc7d6105052c25.png"},{"id":71139880,"identity":"4479d413-bbcb-439f-8263-4ca4cefdf0de","added_by":"auto","created_at":"2024-12-11 13:38:47","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":115959,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eAssociation between BMIgap and weight change\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) correlation between BMIgap and weight change across different age ranges, time-points and clinical groups, (B) relevant features contributing to predict 7% weight gain, and (C) comparison of model performance between for 7% weight gain at T2 while with and without BMIgap as feature.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-5259910/v1/994db541bd12ebcfd42590f5.png"},{"id":93996262,"identity":"26217553-a41f-49f3-87a8-c99e76802bc7","added_by":"auto","created_at":"2025-10-21 07:08:20","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1817191,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5259910/v1/37800dd8-bb6c-4bc8-adad-d29c805dbf88.pdf"},{"id":71140454,"identity":"b7a83194-8ffc-4bb7-bf73-ac79e0fedf2e","added_by":"auto","created_at":"2024-12-11 13:46:47","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":9309703,"visible":true,"origin":"","legend":"Supplementary Material","description":"","filename":"GMVBMIsupplementalmaterialsubmition.docx","url":"https://assets-eu.researchsquare.com/files/rs-5259910/v1/49298394982d360be9983630.docx"}],"financialInterests":"\u003cb\u003eYes\u003c/b\u003e there is potential Competing Interest.\nNK owns issued patent US20160192889A1 (‘Adaptive pattern recognition for psychosis risk modeling’). As a member of Spring Health's Scientific Advisory Board, NK has advised that company on the development of tools to predict treatment outcomes for depression and psychosis. He has no equity and has received no financial compensation from this company. AB reports speaker fees from Otsuka, Lundbeck, Angelini and Rovi outside of the submitted work. JH reports personal fees from Orion ltd, personal fees from Lundbeck, personal fees from Otsuka and others from Takeda during the conduct of the study. RU reports speaker fees from Sunovion, Otsuka and Vitaris outside the submitted work as well as unpaid officership with the British Association for Pharmacology - Honorary General Secretary 2021-2024. She serves as Deputy Editor for The British Journal of Psychiatry. PF reports he has received research support/honoraria for lectures or advisory activities from Boehringer-Ingelheim, Janssen, Lundbeck, Otsuka, Recordati and Richter outside the submitted work. OAA is a consultant to Cortechs.ai and has received speaker’s honorarium from Otsuka, Lundbeck, Janssen and Sunovion. RL reports honoraria for lectures or advisory activities from Boehringer-Ingelheim, Janssen, Otsuka, and ROVI outside the submitted work. The remaining authors have nothing to disclose.","formattedTitle":"BMIgap: a new tool to quantify transdiagnostic brain signatures of current and future weight","fulltext":[{"header":"Introduction","content":"\u003cp\u003eGlobally 26% of adults are overweight (BMI of 25\u0026ndash;30 kg/m\u003csup\u003e2\u003c/sup\u003e) with an additional 13% classified as obese (BMI\u0026thinsp;\u0026ge;\u0026thinsp;30 kg/m\u003csup\u003e2\u003c/sup\u003e) (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e), highlighting the pandemic nature of obesity (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e). Given its strong associations with metabolic diseases, including cardiovascular disease and type 2 diabetes, obesity stands as a major risk factor for somatic disorders (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e). Importantly, obesity frequently parallels psychiatric disorders such as schizophrenia (SCZ), major depressive disorder (MDD), bipolar, personality and anxiety disorders (\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e). Psychiatric patients have a two-to-three-fold higher incidence of obesity and metabolic diseases compared to the general population, significantly contributing to their excess mortality (\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e). Secondary factors such as smoking, alcohol use, sedentary lifestyle, and commonly prescribed medications also significantly impact body weight and metabolic syndromes in psychiatric patients (\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e). Particularly atypical antipsychotics and antidepressants have been implicated in weight gain (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e). Furthermore, individuals exhibiting negative symptoms\u0026mdash;characterized by affective flattening, anhedonia and avolition\u0026mdash;are at an increased risk for weight gain (\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e). In turn, weight gain and obesity can negatively impact an affected individual\u0026rsquo;s quality of life and self-esteem, thereby further exacerbating already pre-existing mental health issues and ultimately leading to a vicious circle between psychiatric symptoms, secondary disease effects and weight gain (\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eIn healthy populations, neuroimaging studies have identified associations between higher BMI and reduced grey matter volume (GMV) in the prefrontal, temporal, parietal and occipital cortices, the cerebellum, insula, thalamus, and amygdala (\u003cspan additionalcitationids=\"CR17\" citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e). These associations might be related to lowered inhibitory control activity for food-related stimuli and responses resulting from a disbalance of brain activity in cognitive control regions and reward-related regions (\u003cspan additionalcitationids=\"CR18\" citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e). These findings suggest a plausible relationship between cognitive control deficits and overeating (\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e) potentially resulting in higher caloric intake and obesity (\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e). Furthermore, in patients with SCZ, higher BMI was found to be associated with reduced GMV in the prefrontal cortex (PFC), specifically the orbitofrontal cortex (OFC), and the hippocampus (\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e). In patients with MDD, higher BMI was correlated with reduced GMV in the medial prefrontal regions, particularly in areas involved in impulse control and emotion regulation (\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eHowever, a comprehensive understanding of the neurobiological underpinnings of obesity, particularly in in psychiatric disorders, is lacking as are methods to identify those at risk for weight gain (\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e, \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e). Therefore, personalized tools are needed to capture these interactions and enable targeted interventions\u0026mdash;such as exercise, psychotherapy, medications, or brain stimulation\u0026mdash;to prevent weight gain, improve treatment adherence, and reduce excess mortality (\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eTo better understand the neurobiological underpinnings of BMI and weight gain in early stages of affective and psychotic disorders, we developed an individualized BMI prediction model based on whole-brain GMV in healthy control (HC) individuals (HC\u003csub\u003ediscovery\u003c/sub\u003e) and validated it in a HC sample (HC\u003csub\u003evalidation\u003c/sub\u003e). We then applied this BMI predictor to individuals suffering from schizophrenia (SCZ), recent-onset depression (ROD), and clinical high-risk states for psychosis (CHR). We introduce BMIgap (Body Mass Index gap estimation) as an individualized brain-based metric, which calculates the difference between brain-estimated and measured BMI (BMI\u003csub\u003epredicted\u003c/sub\u003e-BMI\u003csub\u003emeasured\u003c/sub\u003e). Next, we correlated BMIgap with future weight changes at 1- and 2-years follow-up at the group level and used it along with additional clinical factors to predict future weight gain at the individual level. Finally, we explored the phenotypic association between BMIgap, SCZ expression (i.e. neuroanatomical similarity for SCZ), and clinical features using sparse partial least squares (SPLS). We hypothesized that (i) accurate models for individualized BMI prediction can be derived from structural brain imaging using supervised machine learning, (ii) interactions between BMI-predictive and disease-specific brain signatures result in systematic brain deviations that are captured by BMIgap, (iii) BMIgap is associated with key measures of disease severity, including age of onset, illness duration or hospitalization frequency, and (iv) BMIgap can serve as a personalized brain-based tool to assess future weight gain and identify at-risk individuals in early disease stages.\u003c/p\u003e"},{"header":"Materials and methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\n \u003ch2\u003eParticipants\u003c/h2\u003e\n \u003cp\u003eThe study followed the Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis (TRIPOD) reporting guidelines (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.equator-network.org/reporting-guidelines/tripod-statement/\u003c/span\u003e\u003c/span\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e).\u003c/span\u003e We included T1-weighted MRI scans of 1504 HC individuals (HC\u003csub\u003ediscovery\u003c/sub\u003e=770; HC\u003csub\u003evalidation\u003c/sub\u003e=734) from four independent datasets covering 14 sites: Information eXtraction from Images (IXI; \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://brain-development.org/ixi-dataset/\u003c/span\u003e\u003c/span\u003e), Personalized Prognostic Tools for Early Psychosis Management (PRONIA; \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ewww.pronia.eu\u003c/span\u003e\u003c/span\u003e), Norwegian Centre for Mental Disorders Research (NORMENT; \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.med.uio.no/norment/\u003c/span\u003e\u003c/span\u003e (\u003cspan class=\"CitationRef\"\u003e26\u003c/span\u003e)) and the Munich Brain Imaging Database (MUC, (\u003cspan class=\"CitationRef\"\u003e27\u003c/span\u003e)) (Supplementary Methods). Moreover, three patient populations were included: SCZ individuals (N\u0026thinsp;=\u0026thinsp;146) from the MUC cohort and CHR (N\u0026thinsp;=\u0026thinsp;213) and ROD individuals (N\u0026thinsp;=\u0026thinsp;200) from the PRONIA study (Table \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e, Table S2). All studies were approved by their local ethics committee and adhered to the ethical standards as outlined in the Declaration of Helsinki (\u003cspan class=\"CitationRef\"\u003e28\u003c/span\u003e).\u003c/p\u003e\n \u003cdiv class=\"gridtable\"\u003e\u0026nbsp;\u003ctable id=\"Tab1\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eSociodemographic differences in discovery, validation, and clinical groups.\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003ccolgroup cols=\"10\"\u003e\u003c/colgroup\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eHC\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003et/\u0026chi;\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eP\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\" colspan=\"3\"\u003e\n \u003cp\u003eClinical group\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eF/\u0026chi;\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eP\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eDiscovery\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eValidation\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eSCZ\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eCHR\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eROD\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003cth align=\"left\"\u003e\u0026nbsp;\u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSample (N)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e770\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e734\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.86 \u003csup\u003eb\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.35\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e146\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e213\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e200\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e776.60 \u003csup\u003eb\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;\u0026thinsp;.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBMI, Mean (SD)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e25.10 (4.03)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e23.03 (2.07)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e12.49 \u003csup\u003ea\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;\u0026thinsp;.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e24.02 (3.40)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e23.46 (3.42)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e24.01 (3.57)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e13.48 \u003csup\u003ec\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;\u0026thinsp;.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAge, Mean (SD)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e41.26 (15.51)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e32.24 (12.75)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e12.29 \u003csup\u003ea\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;\u0026thinsp;.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e30.83 (9.97)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e23.92 (5.24)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e26.02 (6.37)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e164.42 \u003csup\u003ec\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;\u0026thinsp;.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSex (Female, N (%))\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e435 (56.49)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e373 (50.82)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e4.75 \u003csup\u003eb\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.029\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e34.00 (23.29)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e103 (48.36)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e96 (48.00)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e55.17 \u003csup\u003eb\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;\u0026thinsp;.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSymptoms, Mean (SD)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePANSS (total)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e52.26 (29.45)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e52.07 (18.22)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e48.44 (14.26)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.08 \u003csup\u003ec\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.13\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePANSS (positive)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e11.92 (8.13)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e11.31 (4.16)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e8.20 (2.21)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e29.09 \u003csup\u003ec\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;\u0026thinsp;.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePANSS (negative)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e14.98 (9.73)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e13.09 (6.86)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e12.50 (5.66)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e5 \u003csup\u003ec\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e.007\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePANSS (general)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e25.36 (16.13)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e27.66 (10.05)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e27.74 (8.84)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.19 \u003csup\u003ec\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e.113\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSANS (total)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e44.44 (26.52)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e27.08 (24.88)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e23.91 (20.95)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e30.69 \u003csup\u003ec\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;\u0026thinsp;.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBDI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3.55 (5.05)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3.27 (5.20)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.50 \u003csup\u003ea\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.616\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e24.09 (11.75)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e24.56 (12.48)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e182.47 \u003csup\u003ec\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;\u0026thinsp;.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eFunctioning, Mean (SD)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eGAF:S past month\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e86.30 (6.43)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e87.32 (6.16)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.54 \u003csup\u003ea\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.124\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e51.31 (11.29)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e53.92 (13.17)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e503.97 \u003csup\u003ec\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;\u0026thinsp;.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eGAF:D past month\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e85.49 (6.32)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e86.51 (5.89)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.59 \u003csup\u003ea\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.113\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e51.94 (12.)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e54.14 (3.60)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e385.24 \u003csup\u003ec\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;\u0026thinsp;.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eGF:S current\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e8.50 (0.89)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e8.54 (0.71)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.51 \u003csup\u003ea\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.612\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e6.25 (1.75)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e6.39 (1.23)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e157.82 \u003csup\u003ec\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;\u0026thinsp;.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eGF:R current\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e8.49 (0.77)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e8.59 (0.67)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.40 \u003csup\u003ea\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.164\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e5.80 (1.70)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e5.97 (1.81)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e150.54 \u003csup\u003ec\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;\u0026thinsp;.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eWHOQoL (total)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e97.45 (33.60)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e102.81 (28.91)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.65 \u003csup\u003ea\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.099\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e71.20 (28.81)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e74.92 (24.86)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e39.58 \u003csup\u003ec\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;\u0026thinsp;.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMedications\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAntipsychotics (N (%))\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e60 (28.17)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e27 (13.50)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e12.51 \u003csup\u003eb\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e\u0026lt;\u0026thinsp;0.001\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAntidepressants (N (%))\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e105 (49.30)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e137 (68.50)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e4.23 \u003csup\u003eb\u003c/sup\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.040\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCPZ-equivalent (mg)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e358.9 (382.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNA\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003ctfoot\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"10\"\u003e\u003csup\u003ea\u003c/sup\u003e \u0026chi;\u003csup\u003e2\u003c/sup\u003e-test, \u003csup\u003eb\u003c/sup\u003e t-test, \u003csup\u003ec\u003c/sup\u003e F-test\u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tfoot\u003e\n \u003c/table\u003e\n \u003c/div\u003e\n \u003cp\u003eWe included HC individuals aged 15\u0026ndash;75 years with a BMI of 18.5\u0026ndash;35 kg/m\u003csup\u003e2\u003c/sup\u003e without current or previous psychiatric disorders, in the discovery model by following these steps: (i) to avoid an underrepresentation of the tails of the BMI distribution and to maximize model generalizability throughout the investigated BMI range, we sampled 770 HC individuals into 33 BMI bins (0.5 BMI per bin) to approximate a uniform BMI distribution (Figure \u003cspan class=\"InternalRef\"\u003eS1\u003c/span\u003ea); (ii) age distribution of individuals was matched within the BMI bins to mitigate the natural correlation between age and BMI (\u003cspan class=\"CitationRef\"\u003e29\u003c/span\u003e) (Figure \u003cspan class=\"InternalRef\"\u003eS1\u003c/span\u003ec). These 770 HC individuals constituted the HC\u003csub\u003ediscovery\u003c/sub\u003e sample, and the remaining 734 HC individuals constituted the HC\u003csub\u003evalidation\u003c/sub\u003e sample. The clinical population consisted of individuals with SCZ (N\u0026thinsp;=\u0026thinsp;146), CHR (N\u0026thinsp;=\u0026thinsp;213) and ROD (N\u0026thinsp;=\u0026thinsp;200) (Figures \u003cspan class=\"InternalRef\"\u003eS1\u003c/span\u003e-S2, Supplementary Methods).\u003c/p\u003e\n\u003c/div\u003e\n\u003ch3\u003eMRI data acquisition and pre-processing\u003c/h3\u003e\n\u003cp\u003eThe Participants in the IXI and PRONIA studies underwent MRI scanning at 1.5T or 3T, while NORMENT and MUC participants were examined at 1.5T MRI scanners (Supplementary Methods, Table \u003cspan class=\"InternalRef\"\u003eS1\u003c/span\u003e). To facilitate between-study comparability, we applied the VBM8 pre-processing pipeline described by Opel et al (\u003cspan class=\"CitationRef\"\u003e18\u003c/span\u003e) to produce normalized, modulated GMV tissue maps (Supplementary Methods). For computational efficiency and noise reduction, GMV images were resliced to a 3x3x3 mm\u003csup\u003e3\u003c/sup\u003e isotropic voxel resolution.\u003c/p\u003e\n\u003ch3\u003eMachine learning analysis\u003c/h3\u003e\n\u003cp\u003eThe open-source machine learning software NeuroMiner \u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e(v1.1;\u003c/span\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/neurominer-git/NeuroMiner_1.1\u003c/span\u003e\u003c/span\u003e\u003cspan type=\"Underline\" class=\"Underline\" name=\"Emphasis\"\u003e)\u003c/span\u003e was used for training and application of all supervised machine learning models. To prevent information leakage between train and test data and enhance model generalizability, we implemented a repeated nested cross-validation (CV) cycle with five folds and permutations each at the inner and outer CV cycles. Thus, the process of model optimization was completely insulated from the process of model validation (Supplementary Methods).\u003c/p\u003e\n\u003cp\u003eData pre-processing included Gaussian smoothing of the neuroimaging data, partial correlation analysis to remove effects of chronological age from the GMV data, global mean offset correction to remove site effects, i.e. MR scanner effects, principal component analysis for dimensionality reduction, and voxel-wise scaling between 0 and 1 (Supplementary Methods). We used 𝜈-support vector machine regression with a linear kernel to predict BMI based on whole-brain voxel-wise GMV (71276 features) using the mean absolute error (MAE) as the optimization criterion. The statistical significance of the model was evaluated using 1000 permutations of the BMI label (\u0026alpha;\u0026thinsp;=\u0026thinsp;0.05). Predictive brain patterns were visualized at the voxel level combining the grand means of the cross-validation ratio (CVR) (\u003cspan class=\"CitationRef\"\u003e30\u003c/span\u003e) and sign-based consistency mapping, thereby assessing whether a feature consistently predicted higher or lower BMI across CV partitions (\u003cspan class=\"CitationRef\"\u003e32\u003c/span\u003e) (Supplementary Methods).\u003c/p\u003e\n\u003cp\u003eWe applied this BMI predictor to the HC\u003csub\u003evalidation\u003c/sub\u003e to assess the model\u0026rsquo;s generalizability. Furthermore, the BMI predictor was applied to SCZ, ROD and CHR individuals to obtain brain-based BMI predictions for these clinical populations. BMIgap was calculated by subtracting the measured from the predicted BMI scores (BMIgap\u0026thinsp;=\u0026thinsp;BMI\u003csub\u003epredicted\u003c/sub\u003e\u0026ndash;BMI\u003csub\u003emeasured\u003c/sub\u003e), where a positive BMIgap reflected a higher brain-based prediction of BMI than the measured BMI, and vice versa. To mitigate systematic bias in predicted BMI, characterized by over-estimation at lower BMI ranges and under-estimation at higher BMI ranges, BMIgap were adjusted for BMI through partial correlation analysis (\u003cspan class=\"CitationRef\"\u003e33\u003c/span\u003e) (Supplementary Methods, Figure S3-S4). We used the corrected BMIgap for all further analysis steps. Furthermore, we independently calculated BMIgap for antipsychotic and antidepressant na\u0026iuml;ve CHR and ROD individuals to investigate whether medication influenced BMIgap in these subpopulations. Since, there were no unmedicated SCZ patients, we correlated BMIgap with their chlorpromazine equivalents.\u003c/p\u003e\n\u003ch3\u003eClinical Investigation of BMIgap\u003c/h3\u003e\n\u003cp\u003eTo understand the clinical implications of BMIgap, we analyzed its relationship with clinical variables, particularly within the SCZ group. First, to identify brain patterns distinguishing patients with SCZ from HC individuals, we classified SCZ vs. HC\u003csub\u003ediscovery\u003c/sub\u003e in the MUC sample using the same pre-processing and CV settings as for the BMI-prediction model. The binarized sign-consistency maps derived from SCZ and BMI predictors were overlapped to identify brain regions commonly predictive of both phenotypes (Supplement methods). Furthermore, we extracted the decision scores from the SCZ classifier, which we refer to as SCZ expression score. A higher SCZ expression score indicates a higher likelihood to be classified as SCZ and therefore a greater neuroanatomical similarity to SCZ, while a lower SCZ expression score reflects higher likelihood of HC classification and neuroanatomical HC similarity.\u003c/p\u003e\n\u003cp\u003eNext, we studied the covariation between BMIgap, SCZ likeness, and clinical variables within the overlapping brain regions of SCZ and BMI. To this end, we employed multivariate SPLS using the SPLS Toolbox by Popovic et al. (\u003cspan class=\"CitationRef\"\u003e34\u003c/span\u003e, \u003cspan class=\"CitationRef\"\u003e35\u003c/span\u003e) to investigate covariance patterns between two data domains in the SCZ sample: (i) a six feature matrix including BMIgap, SCZ expression scores, PANSS total score, age of onset, illness duration and number of hospitalizations and, (ii) a brain data matrix containing the vectorized voxels extracted using the binarized mask of overlapping BMI- and SCZ-predictive voxels. The SPLS algorithm uses singular value decomposition to generate multiple layers of distinct, multivariate associative effects between the two data matrices, called latent variables (LV) (Supplementary Methods).\u003c/p\u003e\n\u003ch3\u003eInvestigation of BMIgap and future weight change\u003c/h3\u003e\n\u003cp\u003eWe investigated the association between BMIgap and future weight change using two approaches: (i) correlation analysis, and (ii) ML-based prediction of weight gain. In the PRONIA cohort, where longitudinal data was available, we correlated the BMIgap to weight changes at 1- (T1) and 2-year (T2) follow-up. Weight changes were calculated as the difference between the weight at the follow-up time point and the weight at baseline (T0): 1-year (\u0026Delta;W\u003csub\u003e1\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;Weight\u003csub\u003eT1\u003c/sub\u003e -Weight\u003csub\u003eT0\u003c/sub\u003e), and 2 years (\u0026Delta;W\u003csub\u003e2\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;Weight\u003csub\u003eT2\u003c/sub\u003e-Weight\u003csub\u003eT0\u003c/sub\u003e). Correlation analyses were conducted for the entire cohort, as well as separately for sex and study group. In the first step, we analyzed the correlation between BMIgap and all observed weight changes. Building on previous literature, we correlated BMIgap to weight changes only within subpopulations of patients that exhibited at least\u0026thinsp;+\u0026thinsp;3%, +\u0026thinsp;5% or +\u0026thinsp;7% weight gain at the respective follow-up (\u003cspan class=\"CitationRef\"\u003e36\u003c/span\u003e\u0026ndash;\u003cspan class=\"CitationRef\"\u003e38\u003c/span\u003e). Additionally, we analyzed these correlations across different age ranges, including individuals over 15, 20, 25, 30, and 35 years, as well as within 5-year bins: 15\u0026ndash;20, 20\u0026ndash;25, 25\u0026ndash;30, 30\u0026ndash;35, and 35\u0026ndash;40 years. Finally, we used the three weight gain thresholds (+\u0026thinsp;3%, +\u0026thinsp;5%, and +\u0026thinsp;7%) as classification criteria to predict whether CHR and ROD individuals experienced weight gain above these thresholds at T1 and T2 or not. We used BMIgap as well as age, sex, study group (ROD, CHR), exercise (strenuous exercise or mindfulness activities such as yoga and meditation), and history of somatic comorbidities (i.e. whether the individual suffered from somatic illness) as features. Classification analyses were conducted with and without BMIgap as a feature to assess if BMIgap significantly impacts the prediction of future weight gain.\u003c/p\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003eIndividualized BMI prediction\u003c/h2\u003e \u003cp\u003eThe model predicted BMI in the HC\u003csub\u003ediscovery\u003c/sub\u003e individuals with an MAE of 2.75 kg/m\u003csup\u003e2\u003c/sup\u003e (R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.28,P\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and generalized to the HC\u003csub\u003evalidation\u003c/sub\u003e with an MAE of 2.29 kg/m\u003csup\u003e2\u003c/sup\u003e for HC (R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.26,P\u0026thinsp;\u0026lt;\u0026thinsp;0.001, Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea). Applied to the clinical subpopulations, the BMI predictor yielded an MAE of 2.85 kg/m\u003csup\u003e2\u003c/sup\u003e for SCZ (R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.25,P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), an MAE of 3.07 kg/m\u003csup\u003e2\u003c/sup\u003e for CHR (R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.16,P\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and an MAE of 2.73 kg/m\u003csup\u003e2\u003c/sup\u003e for ROD (R\u003csup\u003e2\u003c/sup\u003e\u0026thinsp;=\u0026thinsp;0.10,P\u0026thinsp;\u0026lt;\u0026thinsp;0.001) individuals (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e; Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eb).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eModel performances of the regression analysis for the discovery model and its application to the replication and patient groups.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"11\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c10\" colnum=\"10\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c11\" colnum=\"11\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStudy group\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eN\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eBMIgap uncorrected (kg/m\u003csup\u003e2\u003c/sup\u003e)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003eBMIgap\u003c/p\u003e \u003cp\u003e(kg/m\u003csup\u003e2\u003c/sup\u003e)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e \u003cp\u003eMAE\u003c/p\u003e \u003cp\u003e(kg/m\u003csup\u003e2\u003c/sup\u003e)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c9\" namest=\"c8\"\u003e \u003cp\u003eR\u003csup\u003e2\u003c/sup\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c10\"\u003e \u003cp\u003e\u003cem\u003er\u003c/em\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c11\"\u003e \u003cp\u003eP\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHC\u003csub\u003ediscovery\u003c/sub\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e770\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.01 (3.4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0 (1.78)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e2.75\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e0.28\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c10\" namest=\"c9\"\u003e \u003cp\u003e0.53\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;0.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHC\u003csub\u003evalidation\u003c/sub\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e734\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.73 (2.2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.23 (1.68)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e2.29\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e0.26\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c10\" namest=\"c9\"\u003e \u003cp\u003e0.51\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;0.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSCZ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e146\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.83 (3.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1.05 (1.53)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e2.85\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e0.25\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c10\" namest=\"c9\"\u003e \u003cp\u003e0.50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;0.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCHR\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e213\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e1.70 (3.26)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.51 (1.68)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e3.07\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e0.16\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c10\" namest=\"c9\"\u003e \u003cp\u003e0.40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e\u003cb\u003e\u0026lt;\u0026thinsp;0.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eROD\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e200\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e-0.03 (3.48)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-0.82 (1.64)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003e2.73\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003e0.10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c10\" namest=\"c9\"\u003e \u003cp\u003e0.32\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c11\"\u003e \u003cp\u003e\u003cb\u003e0.001\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eLower GMV in the cerebellar, prefrontal, occipital, and insular cortices, the postcentral gyrus, hippocampus, and thalamus was predictive of higher BMI. Lower GMV in the left hemisphere involving the cingulate, cerebellar, inferior occipital and temporal cortices, as well as in the right hemisphere covering parts of the precuneus, putamen and Rolandic operculum were predictive of lower BMI (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ec, Figure S5, Table S3).\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eBMIgap estimation across clinical groups\u003c/h3\u003e\n\u003cp\u003eThe application of the BMI predictor yielded a mean (\u0026plusmn;\u0026thinsp;SD) BMIgap of 0.23 (\u0026plusmn;\u0026thinsp;1.68) kg/m\u003csup\u003e2\u003c/sup\u003e for HC\u003csub\u003evalidation\u003c/sub\u003e, 1.05 (\u0026plusmn;\u0026thinsp;1.53) kg/m\u003csup\u003e2\u003c/sup\u003e for SCZ, -0.82 (\u0026plusmn;\u0026thinsp;1.64) kg/m\u003csup\u003e2\u003c/sup\u003e for ROD and 0.51 (\u0026plusmn;\u0026thinsp;1.68) kg/m\u003csup\u003e2\u003c/sup\u003e for CHR individuals. BMIgap differed between HC\u003csub\u003ediscovery\u003c/sub\u003e and HC\u003csub\u003evalidation\u003c/sub\u003e individuals and clinical groups (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ed; F\u003csub\u003e[discovery vs. patients]\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;33.90, P\u0026thinsp;\u0026lt;\u0026thinsp;0.001; F\u003csub\u003e[validation vs. patients]\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;32.36, P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Notably, post-hoc pairwise comparisons revealed significant differences in BMIgap between the HC\u003csub\u003ediscovery\u003c/sub\u003e and HC\u003csub\u003evalidation\u003c/sub\u003e groups (t\u0026thinsp;=\u0026thinsp;2.62, P\u0026thinsp;=\u0026thinsp;0.009) likely due to variations in BMI distributions, as indicated by a significant difference in variances (F\u0026thinsp;=\u0026thinsp;3.8018, P\u0026thinsp;\u0026lt;\u0026thinsp;0.001; Figure \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003eA,C). BMIgap differed significantly between HC\u003csub\u003ediscovery\u003c/sub\u003e and all clinical groups with the highest BMIgap for SCZ (t\u0026thinsp;=\u0026thinsp;6.68,P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), followed by CHR (t\u0026thinsp;=\u0026thinsp;3.72,P\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and the lowest for ROD (t=-5.88,P\u0026thinsp;\u0026lt;\u0026thinsp;0.001).\u003c/p\u003e \u003cp\u003eWe did not find significant differences in BMIgap between medication na\u0026iuml;ve (N\u0026thinsp;=\u0026thinsp;80 (37.56%)), antipsychotic-na\u0026iuml;ve (N\u0026thinsp;=\u0026thinsp;153 (71.83%)), antidepressant-na\u0026iuml;ve (N\u0026thinsp;=\u0026thinsp;108 (50.70%)) and concurrently antidepressant-antipsychotics-treated (N\u0026thinsp;=\u0026thinsp;133 (62.44%)) within the CHR individuals (F\u0026thinsp;=\u0026thinsp;0.6, P\u0026thinsp;=\u0026thinsp;0.6244). Furthermore, we did not find significant differences between BMIgap of medication na\u0026iuml;ve (N\u0026thinsp;=\u0026thinsp;59 (29.50%)), antipsychotic-na\u0026iuml;ve (N\u0026thinsp;=\u0026thinsp;173 (86.50%)), antidepressant-na\u0026iuml;ve (N\u0026thinsp;=\u0026thinsp;63 (31.50%)) and concurrently antidepressant-antipsychotics-treated (N\u0026thinsp;=\u0026thinsp;141 (70.50%)) ROD individuals (F\u0026thinsp;=\u0026thinsp;0.002, P\u0026thinsp;=\u0026thinsp;0.9964). Moreover, we did not find a significant correlation between BMIgap and chlorpromazine equivalents (\u003cem\u003er\u003c/em\u003e=-0.01,P\u0026thinsp;=\u0026thinsp;0.86) in the SCZ sample.\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eSCZ specific brain-signatures\u003c/h2\u003e \u003cp\u003eThe SCZ classifier yielded a BAC of 72.4% (sensitivity\u0026thinsp;=\u0026thinsp;72.2%, specificity\u0026thinsp;=\u0026thinsp;72.6%; P\u0026thinsp;\u0026lt;\u0026thinsp;0.001) in separating HC from SCZ individuals. Voxels predictive of SCZ were found predominantly in the inferior, middle and superior frontal gyrus, as well as in hippocampal, thalamic, insular, Rolandic operculum, postcentral, cerebellar, and basal ganglia structures. On the right hemisphere the lingual, fusiform gyrus and mid-temporal lobe were predictive of the SCZ class membership (Figure S6). The brain patterns of the BMI predictor and the SCZ classifier overlapped in the inferior, middle and superior frontal gyrus, caudate, putamen, Rolandic operculum, right precuneus and the middle temporal lobe regions (Figure S7).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eClinical associations of shared SCZ and BMI signatures\u003c/h2\u003e \u003cp\u003eThe SPLS analysis yielded five significant LVs, representing distinct levels of association between the neuroanatomic overlap regions of the BMI and SCZ models and the clinical disease features (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, Figure S8). While LV2, LV3, and LV5 captured disease-specific patterns, LV1 and LV4 extracted covariate patterns of age and sex (Supplement Results).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eIn LV2 (\u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.84, P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), higher BMIgap, SCZ expression scores, age of onset, number of hospitalizations and illness duration were related to decreased GMV in the default mode network (DMN) specifically in the A, B and C subcomponents, visual, somatomotor-A, attention, salience, limbic, and control networks and increased GMV in DMN (auditory) and somatomotor B networks (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea).\u003c/p\u003e \u003cp\u003eIn LV3 (\u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.85, P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), higher BMIgap and higher SCZ expression scores were related to decreased GMV in the DMN-B and increased GMV in the DMN-D (auditory) (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb).\u003c/p\u003e \u003cp\u003eIn LV5 (\u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.58, P\u0026thinsp;\u0026lt;\u0026thinsp;0.001), lower PANSS total score, illness duration, and age of onset and higher SCZ expression score were related to decreased GMV in DMN-C and Control-C networks, and increased GMV in the DMN-A/B, Somatomotor-B, Dorsal attention-B, and salience networks (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ec).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eBMIgap and future weight change\u003c/h2\u003e \u003cp\u003eIn HC individuals, BMIgap showed a positive correlation with 2-year weight gain (ΔW\u003csub\u003e2\u003c/sub\u003e, N\u0026thinsp;=\u0026thinsp;216; \u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.14,P\u0026thinsp;\u0026lt;\u0026thinsp;0.05). Specifically, at T2, adults between in the age 25\u0026ndash;40 years subgroup (N\u0026thinsp;=\u0026thinsp;46; \u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.46,P\u0026thinsp;\u0026lt;\u0026thinsp;0.001). BMIgap of ROD individuals positively correlated with both ΔW\u003csub\u003e1\u003c/sub\u003e (N\u0026thinsp;=\u0026thinsp;141; \u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.18,P\u0026thinsp;\u0026lt;\u0026thinsp;0.05) and ΔW\u003csub\u003e2\u003c/sub\u003e (N\u0026thinsp;=\u0026thinsp;92; \u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.3,P\u0026thinsp;\u0026lt;\u0026thinsp;0.01) at all ages and notably for young individuals between 15\u0026ndash;20 years (ΔW\u003csub\u003e2\u003c/sub\u003e:N\u0026thinsp;=\u0026thinsp;15; \u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.52,P\u0026thinsp;\u0026lt;\u0026thinsp;0.05). Moreover, female ROD individuals showed a significant correlation between BMIgap and weight gain at T1 (ΔW\u003csub\u003e1\u003c/sub\u003e:N\u0026thinsp;=\u0026thinsp;70; \u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.29,P\u0026thinsp;\u0026lt;\u0026thinsp;0.05), while male individuals did not (N\u0026thinsp;=\u0026thinsp;71; \u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.02,P\u0026thinsp;=\u0026thinsp;0.86). CHR individuals between 20\u0026ndash;40 years of age showed significant correlations between BMIgap and ΔW\u003csub\u003e1\u003c/sub\u003e in the +\u0026thinsp;5% weight increase subgroup (N\u0026thinsp;=\u0026thinsp;58; \u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.20,P\u0026thinsp;\u0026lt;\u0026thinsp;0.05) as well as ΔW\u003csub\u003e2\u003c/sub\u003e in the +\u0026thinsp;3% (N\u0026thinsp;=\u0026thinsp;39; \u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.28,P\u0026thinsp;\u0026lt;\u0026thinsp;0.05) and +\u0026thinsp;5% (N\u0026thinsp;=\u0026thinsp;32; \u003cem\u003er\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.36,P\u0026thinsp;\u0026lt;\u0026thinsp;0.05) subgroup (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA, Figure S10).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eWeight gain predictor\u003c/h2\u003e \u003cp\u003eThe multivariate weight gain prediction model indicated a weight gain of +\u0026thinsp;7% at T2 with a BAC of 59.2% (sensitivity\u0026thinsp;=\u0026thinsp;64.9%, specificity\u0026thinsp;=\u0026thinsp;53.5%, P\u0026thinsp;\u0026lt;\u0026thinsp;0.05), while the +\u0026thinsp;3% (BAC\u0026thinsp;=\u0026thinsp;52.2%; sensitivity\u0026thinsp;=\u0026thinsp;52.5%, specificity\u0026thinsp;=\u0026thinsp;51.9%, P\u0026thinsp;=\u0026thinsp;0.21) and +\u0026thinsp;5% weight gain prediction models did not perform above chance level (BAC\u0026thinsp;=\u0026thinsp;45.2%, sensitivity\u0026thinsp;=\u0026thinsp;62.5%, specificity\u0026thinsp;=\u0026thinsp;35.2%) (Table S6). Key predictive features for the 7% weight gain predictor included age, clinical group (ROD, CHR), exercise, somatic history, BMI gap, and tobacco use. Specifically, amongst CHR individuals, a higher BMIgap, in combination with a history of somatic conditions and reduced mindfulness-based exercises was predictive of weight gain. For ROD individuals, the BMIgap was negatively associated with 7% weight gain, particularly in individuals with history of lower somatic comorbidities, reduced tobacco use, and higher age (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eC). Notably, the predictive performance of the +\u0026thinsp;7% weight reduced after excluding BMIgap as a feature (BAC\u0026thinsp;=\u0026thinsp;56.1%; sensitivity\u0026thinsp;=\u0026thinsp;48.6%, specificity\u0026thinsp;=\u0026thinsp;63.6%; P\u0026thinsp;\u0026lt;\u0026thinsp;0.05) and these two models differed significantly (P\u0026thinsp;\u0026lt;\u0026thinsp;0.05) (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eC).\u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eThe aim of our study was to introduce BMIgap as a metric to evaluate BMI-related brain signatures, interrogate its overlaps with SCZ brain patterns and explore its implications for future weight gain.\u003c/p\u003e \u003cp\u003eThe BMI predictor relied on lower GMV in frontal and temporal regions and higher GMV in precuneus, putamen and Rolandic operculum to predict BMI. Individuals in the psychosis spectrum (SCZ and CHR) showed an increased BMIgap while individuals suffering from affective diseases (ROD) showed a decreased BMIgap. Moreover, we found that separability of SCZ from HC individuals partly overlapped with BMI-related structural brain variation pertaining to the inhibitory control and reward systems. SPLS analysis revealed that a prefronto-temporal brain pattern predicting both disease and BMI phenotypes was associated with longer illness duration, later disease onset, and higher number of hospitalizations. The potential clinical utility of increased BMIgap, i.e., having a higher brain-estimated BMI than currently measured, was elucidated by our post-hoc analyses. On a group level, higher BMIgap was correlated with future weight gain, an effect particularly pronounced in longer-term trajectories of younger individuals, who were suffering from depressive disorders. On an individual level, BMIgap was identified as a predictive feature for future weight gain in combination with age, clinical group (ROD and CHR), exercise, history of somatic comorbidities, and tobacco usage.\u003c/p\u003e \u003cp\u003eThe co-occurrence of higher BMI and lower GMV in the reward and salience systems in the BMI predictor model may represent a neurobiological mediator of eating behaviors (\u003cspan additionalcitationids=\"CR17 CR18\" citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e). Furthermore, BMI-predictive GMV reductions in brain areas related to taste, reward, and inhibitory control may contribute to increased susceptibility to hypercaloric eating (\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e). In patient groups, the SCZ signature and the BMI signatures overlapped within inhibitory, reward and cognitive control regions, thereby pointing to interacting pathways between obesity and SCZ consistent with previous research (\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e). Our findings substantiate previous reports that prefrontal deficits might lead to reduced cognitive control and therefore may amplify the risk of addictive behaviors such as overeating which could significantly contribute to an increased BMI in psychiatric patients (\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e, \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e, \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e).\u003c/p\u003e \u003cp\u003ePositive BMIgap in SCZ and CHR groups indicate that these patients have brain alterations typically found in healthy individuals with higher BMI. This may imply that underlying pathophysiological processes\u0026mdash;such as inflammation, insulin resistance, and disruptions in gut microbiota\u0026mdash;may influence brain structure similarly to the effects seen in higher BMI conditions and ultimately predispose them to future weight gain (\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e, \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e). These interaction effects may be more pronounced in SCZ compared to CHR individuals given their more advanced disease stage, resulting in higher BMIgap and a higher propensity towards unfavorable metabolic outcomes. Conversely, a negative BMIgap in the ROD group resembles brain signatures found in individuals with a lower BMI. This is supported by previous findings showing overlapping brain patterns between depressed and underweight individuals as well as patients with comorbid depression and anorexia nervosa (\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eMoreover, occurrence of metabolic syndromes in psychiatric patients is often linked to psychiatric medication, namely antipsychotics and antidepressants (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e, \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e). Antipsychotics impact the mesolimbic dopaminergic system and the ventromedial nucleus, thereby altering behavioral responses to environmental stimuli and regulating both food intake and bodyweight (\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e). While antidepressants impact metabolic risk by increasing appetite and suppressing satiety, primarily through histaminergic and serotonin receptor antagonism (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e). However, we did not find significant differences when comparing the BMIgap of fully na\u0026iuml;ve, partly na\u0026iuml;ve and concurrently antidepressant-antipsychotics-treated CHR or ROD. Moreover, we did not find any significant correlations between BMIgap and antipsychotic dosage in SCZ patients. These findings indicate that the BMIgap captures a neurobiological BMI signature, which may represent a more disease-specific or individual predisposition toward future weight changes. This finding aligns with previous research indicating that patients with psychiatric disorders, are at an increased risk of developing metabolic syndrome and obesity, independent of medication use (\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e, \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e), contrasting other literature that primarily attributes weight gain to the effects of psychiatric medication (\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe phenotypic association analysis linked control and reward brain networks to SCZ diagnosis, BMIgap and clinical variables in unique ways. The multivariate signatures of concurrently high BMIgap and SCZ expression scores (LV2, LV3) were associated with both a decrease in GMV within the limbic network and increased GMV within the DMN (auditory) network. In obesity research, these networks have been involved in reward processing (\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e), food motivation (\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e), executive and affective control (\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e), while in SCZ research, they have been particularly implicated in the impaired processing of negative emotions (\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e). Furthermore, later disease onset, higher hospitalization frequency, and longer illness duration, were associated with both higher BMIgap and SCZ-expression scores, thereby implying a potential association between the severity of SCZ and the presence of obesity risk traits. Moreover, we identified a pattern independent of BMIgap (LV5) suggesting that there are distinct effects of SCZ and BMI on the brain, indicating that those with early onset, shorter illness duration, and milder symptoms are less likely to show high SCZ-diagnostic separability, thus, highlighting the complexity of SCZ subtypes and the variability in its manifestation and progression.\u003c/p\u003e \u003cp\u003eMoreover, our assertion that the BMIgap is associated with future weight changes was supported by the positive correlation between BMIgap and future weight change observed across all groups, with the strength of this correlation increasing from 1 to 2-year follow-up. Notably, these correlations were most pronounced among young individuals (15\u0026ndash;20 years) in the ROD group and young adults (25\u0026ndash;40 years) in the CHR group, indicating that individuals in the early stages of psychiatric illness may exhibit distinct brain signatures that predispose them to future weight gain. Furthermore, the 7% weight gain predictor, highlighted that ROD young adults who have higher exercise rates, use less tobacco, have less somatic comorbidities, and a lower BMIgap are less likely to experience weight gain. Conversely, younger CHR individuals with reduced exercise rates, more history of somatic comorbidities, higher tobacco use, and a higher BMIgap were at an increased risk for future weight gain. Notably, the exclusion of the BMIgap feature from the 7% weight gain predictor resulted in a significant reduction of predictive accuracy of the model in correctly predicting future weight gain. Therefore, the BMIgap may be helpful in identifying patients at risk of weight gain, facilitating the early implementation of tailored interventions such as weight-neutral medications, individualized dietary and physical activity regimens, to stabilize weight (\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e).\u003c/p\u003e \u003cdiv id=\"Sec16\" class=\"Section2\"\u003e \u003ch2\u003eLimitations\u003c/h2\u003e \u003cp\u003eAmong the limitations of our study is the lack of individuals with very high (\u0026gt;\u0026thinsp;35) or low (\u0026lt;\u0026thinsp;18) BMI, which should be further addressed by future studies focusing on the understudied subpopulations of underweight and highly obese individuals. Moreover, the study lacked metabolic markers necessary to further interrogate the shared pathophysiological substrates affecting brain structure and clinical covariations between obesity and psychiatric disorders. However, BMI, being widely used in similar studies, enabled us to make preliminary and direct comparisons with prior findings (\u003cspan additionalcitationids=\"CR17\" citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e, \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e, \u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e"},{"header":"Conclusion","content":"\u003cp\u003eIn conclusion, our study identified BMIgap as a crucial metric for exploring the relationship between BMI and brain structure in psychiatric populations, particularly within the depression and schizophrenia spectrum. We found that elevated BMIgap was associated with significant neurobiological alterations in reward and inhibitory control systems, indicating a complex interplay between obesity and schizophrenia. Our findings further suggest that BMIgap could potentially be a predictive indicator of future weight gain, especially among younger individuals with a higher disease burden. Therefore, BMIgap could serve as a template of using machine learning and brain imaging to enhance early identification of patients at risk for metabolic complications. Integrating BMIgap, or future, more sophisticated tools, into clinical assessments may improve strategies for preventing weight gain in psychiatric patients.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions:\u0026nbsp;\u003c/strong\u003eAK, AW and NK had access to all the study data. NK and ESc conceived and designed the study. AK, DP, AW, and NK were responsible for the analytic framework and data interpretation. AK was responsible for data organization. MLP provided the NORMENT data. NK, JK, RKRS, JH, RL, AB, PB, RU, SB, SW, RL, PF, LW, OA and EM were responsible for the data collection. AK, AW, DP and ESa were responsible for the analysis set-up. AK wrote the first manuscript draft critically revised by DP, AW, and NK. ESc, RU, RL, LW, OAA, MB and ES critically revised the manuscript for important intellectual content. All authors read and approved the final manuscript.\u003c/p\u003e\n\u003cp\u003eWe would like to thank Susanne Miedl for helping with the administrative work.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding:\u0026nbsp;\u003c/strong\u003eThis study was supported by the German Federal Ministry of Education and Research (BMBF, 01ZX1904E and 01ZX2204A) as a part of COMorbidity Modeling via Integrative Transfer machine learning in MENTal illness project.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003ePRONIA is a collaboration project funded by the European Union under the 7th Framework Programme under grant agreement no. 602152.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eNK is supported through grants from NIH (U01MH124639-01; ProNET), the Wellcome Trust, the German Innovation Fund (CARE project), the German Federal Ministry of Education and Research (COMMITMENT and BEST projects), as well as ERA PerMed (IMPLEMENT project). AK is funded through the COMMITMENT project. DP was supported by the Else-Kr\u0026ouml;ner-Fresenius-Foundation through the Clinician Scientist Program \u0026lsquo;EKFS-Translational Psychiatry\u0026rsquo;. ESc is supported through grants by the German Federal Ministry of Education and Research (COMMITMENT (grant 01ZX2204A), BEST (grant 01EK2101B), and IMPLEMENT (grant 01KU1905A) projects). The Norwegian study group is funded by the Research Council of Norway and the KG Jebsen stiftelsen.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eRole of funders:\u0026nbsp;\u003c/strong\u003eThe funders were not involved in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.\u003cstrong\u003e\u003cbr\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting Interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eNK owns issued patent US20160192889A1 (\u0026lsquo;Adaptive pattern recognition for psychosis risk modeling\u0026rsquo;). As a member of Spring Health\u0026apos;s Scientific Advisory Board, NK has advised that company on the development of tools to predict treatment outcomes for depression and psychosis. He has no equity and has received no financial compensation from this company. AB reports speaker fees from Otsuka, Lundbeck, Angelini and Rovi outside of the submitted work. JH reports personal fees from Orion ltd, personal fees from Lundbeck, personal fees from Otsuka and others from Takeda during the conduct of the study. RU reports speaker fees from Sunovion, Otsuka and Vitaris outside the submitted work as well as unpaid officership with the British Association for Pharmacology - Honorary General Secretary 2021-2024. She serves as Deputy Editor for The British Journal of Psychiatry. PF reports he has received research support/honoraria for lectures or advisory activities from Boehringer-Ingelheim, Janssen, Lundbeck, Otsuka, Recordati and Richter outside the submitted work. OAA is a consultant to Cortechs.ai and has received speaker\u0026rsquo;s honorarium from Otsuka, Lundbeck, Janssen and Sunovion. RL reports honoraria for lectures or advisory activities from Boehringer-Ingelheim, Janssen, Otsuka, and ROVI outside the submitted work. The remaining authors have nothing to disclose.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eOrganization WH (2005): \u003cem\u003eSurveillance of Chronic Disease Risk Factors : Country Level Data and Comparable Estimates\u003c/em\u003e. World Health Organization. Retrieved May 19, 2024, from https://iris.who.int/handle/10665/43190\u003c/li\u003e\n \u003cli\u003eHeymsfield SB, Wadden TA (2017): Mechanisms, Pathophysiology, and Management of Obesity. \u003cem\u003eN Engl J Med\u003c/em\u003e 376: 254\u0026ndash;266.\u003c/li\u003e\n \u003cli\u003eWeiss F, Barbuti M, Carignani G, Calderone A, Santini F, Maremmani I, Perugi G (2020): Psychiatric Aspects of Obesity: A Narrative Review of Pathophysiology and Psychopathology. \u003cem\u003eJ Clin Med\u003c/em\u003e 9: 2344.\u003c/li\u003e\n \u003cli\u003eKhafagy R, Dash S (2021): Obesity and Cardiovascular Disease: The Emerging Role of Inflammation. \u003cem\u003eFront Cardiovasc Med\u003c/em\u003e 8: 768119.\u003c/li\u003e\n \u003cli\u003eRajan T, Menon V (2017): Psychiatric disorders and obesity: A review of association studies. \u003cem\u003eJ Postgrad Med\u003c/em\u003e 63: 182\u0026ndash;190.\u003c/li\u003e\n \u003cli\u003eAfzal M, Siddiqi N, Ahmad B, Afsheen N, Aslam F, Ali A, \u003cem\u003eet al.\u003c/em\u003e (2021): Prevalence of Overweight and Obesity in People With Severe Mental Illness: Systematic Review and Meta-Analysis. \u003cem\u003eFront Endocrinol\u003c/em\u003e 12. Retrieved March 3, 2023, from https://www.frontiersin.org/articles/10.3389/fendo.2021.769309\u003c/li\u003e\n \u003cli\u003eAbdullah HM, Shahul HA, Hwang MY, Ferrando S (2020): Comorbidity in Schizophrenia: Conceptual Issues and Clinical Management. \u003cem\u003eFocus J Life Long Learn Psychiatry\u003c/em\u003e 18: 386.\u003c/li\u003e\n \u003cli\u003eDe Micheli A, Provenzani U, Krakowski K, Oliver D, Damiani S, Brondino N, \u003cem\u003eet al.\u003c/em\u003e (2024): Physical Health and Transition to Psychosis in People at Clinical High Risk [no. 3]. \u003cem\u003eBiomedicines\u003c/em\u003e 12: 523.\u003c/li\u003e\n \u003cli\u003eKar N, Barreto S (2024): Influence of Lifestyle Factors on Metabolic Syndrome in Psychiatric Patients Attending a Community Mental Health Setting: A Cross-sectional Study. \u003cem\u003eIndian J Psychol Med\u003c/em\u003e 46: 313\u0026ndash;322.\u003c/li\u003e\n \u003cli\u003eSadeghirad B, Duhaney T, Motaghipisheh S, Campbell NRC, Johnston BC (2016): Influence of unhealthy food and beverage marketing on children\u0026rsquo;s dietary intake and preference: a systematic review and meta-analysis of randomized trials. \u003cem\u003eObes Rev Off J Int Assoc Study Obes\u003c/em\u003e 17: 945\u0026ndash;959.\u003c/li\u003e\n \u003cli\u003eH G, B G, S E-H, D C-L, O L, Jd R, \u003cem\u003eet al.\u003c/em\u003e (2020): Antidepressant Medications and Weight Change: A Narrative Review. \u003cem\u003eObes Silver Spring Md\u003c/em\u003e 28. https://doi.org/10.1002/oby.22969\u003c/li\u003e\n \u003cli\u003eA S, L M (2010): Antidepressants and body weight: a comprehensive review and meta-analysis. \u003cem\u003eJ Clin Psychiatry\u003c/em\u003e 71. https://doi.org/10.4088/JCP.09r05346blu\u003c/li\u003e\n \u003cli\u003eHryhorczuk C, Sharma S, Fulton SE (2013): Metabolic disturbances connecting obesity and depression. \u003cem\u003eFront Neurosci\u003c/em\u003e 7: 177.\u003c/li\u003e\n \u003cli\u003eSarwer DB, Polonsky HM (2016): The Psychosocial Burden of Obesity. \u003cem\u003eEndocrinol Metab Clin North Am\u003c/em\u003e 45: 677\u0026ndash;688.\u003c/li\u003e\n \u003cli\u003eLee K, Akinola A, Abraham S (n.d.): Antipsychotic-induced weight gain: exploring the role of psychiatrists in managing patients\u0026rsquo; physical health \u0026ndash; challenges, current options and direction for future care. \u003cem\u003eBJPsych Bull\u003c/em\u003e 48: 24\u0026ndash;29.\u003c/li\u003e\n \u003cli\u003eLi L, Yu H, Zhong M, Liu S, Wei W, Meng Y, \u003cem\u003eet al.\u003c/em\u003e (2022): Gray matter volume alterations in subjects with overweight and obesity: Evidence from a voxel-based meta-analysis. \u003cem\u003eFront Psychiatry\u003c/em\u003e 13. Retrieved March 3, 2023, from https://www.frontiersin.org/articles/10.3389/fpsyt.2022.955741\u003c/li\u003e\n \u003cli\u003eFern\u0026aacute;ndez-And\u0026uacute;jar M, Morales-Garc\u0026iacute;a E, Garc\u0026iacute;a-Casares N (2021): Obesity and Gray Matter Volume Assessed by Neuroimaging: A Systematic Review. \u003cem\u003eBrain Sci\u003c/em\u003e 11: 999.\u003c/li\u003e\n \u003cli\u003eOpel N, Redlich R, Kaehler C, Grotegerd D, Dohm K, Heindel W, \u003cem\u003eet al.\u003c/em\u003e (2017): Prefrontal gray matter volume mediates genetic risks for obesity. \u003cem\u003eMol Psychiatry\u003c/em\u003e 22: 703\u0026ndash;710.\u003c/li\u003e\n \u003cli\u003eStice E, Figlewicz DP, Gosnell BA, Levine AS, Pratt WE (2013): The contribution of brain reward circuits to the obesity epidemic. \u003cem\u003eNeurosci Biobehav Rev\u003c/em\u003e 37: 10.1016/j.neubiorev.2012.12.001.\u003c/li\u003e\n \u003cli\u003eTsai S-Y, Sajatovic M, Hsu J-L, Chung K-H, Chen P-H, Huang Y-J (2020): Body mass index, residual psychotic symptoms, and inflammation associated with brain volume reduction in older patients with schizophrenia. \u003cem\u003eInt J Geriatr Psychiatry\u003c/em\u003e 35: 728\u0026ndash;736.\u003c/li\u003e\n \u003cli\u003eOpel N, Thalamuthu A, Milaneschi Y, Grotegerd D, Flint C, Leenings R, \u003cem\u003eet al.\u003c/em\u003e (2021): Brain structural abnormalities in obesity: relation to age, genetic risk, and common psychiatric disorders. \u003cem\u003eMol Psychiatry\u003c/em\u003e 26: 4839\u0026ndash;4852.\u003c/li\u003e\n \u003cli\u003eHinchliffe N, Capehorn MS, Bewick M, Feenie J (2022): The Potential Role of Digital Health in Obesity Care. \u003cem\u003eAdv Ther\u003c/em\u003e 39: 4397\u0026ndash;4412.\u003c/li\u003e\n \u003cli\u003eBays HE, Fitch A, Cuda S, Gonsahn-Bollie S, Rickey E, Hablutzel J, \u003cem\u003eet al.\u003c/em\u003e (2023): Artificial intelligence and obesity management: An Obesity Medicine Association (OMA) Clinical Practice Statement (CPS) 2023. \u003cem\u003eObes Pillars\u003c/em\u003e 6: 100065.\u003c/li\u003e\n \u003cli\u003eTian YE, Di Biase MA, Mosley PE, Lupton MK, Xia Y, Fripp J, \u003cem\u003eet al.\u003c/em\u003e (2023): Evaluation of Brain-Body Health in Individuals With Common Neuropsychiatric Disorders. \u003cem\u003eJAMA Psychiatry\u003c/em\u003e 80: 567\u0026ndash;576.\u003c/li\u003e\n \u003cli\u003eAlhindi YA, Khalifa N, Al-Khyatt W, Idris I (2023): The use of non-invasive brain stimulation techniques to reduce body weight and food cravings: A systematic review and meta-analysis. \u003cem\u003eClin Obes\u003c/em\u003e 13: e12611.\u003c/li\u003e\n \u003cli\u003eWolfers T, Doan NT, Kaufmann T, Aln\u0026aelig;s D, Moberget T, Agartz I, \u003cem\u003eet al.\u003c/em\u003e (2018): Mapping the Heterogeneous Phenotype of Schizophrenia and Bipolar Disorder Using Normative Models. \u003cem\u003eJAMA Psychiatry\u003c/em\u003e 75: 1146\u0026ndash;1155.\u003c/li\u003e\n \u003cli\u003eKoutsouleris N, Meisenzahl EM, Borgwardt S, Riecher-R\u0026ouml;ssler A, Frodl T, Kambeitz J, \u003cem\u003eet al.\u003c/em\u003e (2015): Individualized differential diagnosis of schizophrenia and mood disorders using neuroanatomical biomarkers. \u003cem\u003eBrain J Neurol\u003c/em\u003e 138: 2059\u0026ndash;2073.\u003c/li\u003e\n \u003cli\u003eGoodyear MDE, Krleza-Jeric K, Lemmens T (2007): The Declaration of Helsinki. \u003cem\u003eBMJ\u003c/em\u003e 335: 624\u0026ndash;625.\u003c/li\u003e\n \u003cli\u003eStevens J, Cai J, Pamuk ER, Williamson DF, Thun MJ, Wood JL (1998): The Effect of Age on the Association between Body-Mass Index and Mortality. \u003cem\u003eN Engl J Med\u003c/em\u003e 338: 1\u0026ndash;7.\u003c/li\u003e\n \u003cli\u003eKoutsouleris N, Kambeitz-Ilankovic L, Ruhrmann S, Rosen M, Ruef A, Dwyer DB, \u003cem\u003eet al.\u003c/em\u003e (2018): Prediction Models of Functional Outcomes for Individuals in the Clinical High-Risk State for Psychosis or With Recent-Onset Depression: A Multimodal, Multisite Machine Learning Analysis. \u003cem\u003eJAMA Psychiatry\u003c/em\u003e 75: 1156\u0026ndash;1172.\u003c/li\u003e\n \u003cli\u003eKoutsouleris N, Dwyer DB, Degenhardt F, Maj C, Urquijo-Castro MF, Sanfelici R, \u003cem\u003eet al.\u003c/em\u003e (2021): Multimodal Machine Learning Workflows for Prediction of Psychosis in Patients With Clinical High-Risk Syndromes and Recent-Onset Depression. \u003cem\u003eJAMA Psychiatry\u003c/em\u003e 78: 195\u0026ndash;209.\u003c/li\u003e\n \u003cli\u003eG\u0026oacute;mez-Verdejo V, Parrado-Hern\u0026aacute;ndez E, Tohka J, Alzheimer\u0026rsquo;s Disease Neuroimaging Initiative (2019): Sign-Consistency Based Variable Importance for Machine Learning in Brain Imaging. \u003cem\u003eNeuroinformatics\u003c/em\u003e 17: 593\u0026ndash;609.\u003c/li\u003e\n \u003cli\u003eZhang B, Zhang S, Feng J, Zhang S (2023): Age-level bias correction in brain age prediction. \u003cem\u003eNeuroImage Clin\u003c/em\u003e 37: 103319.\u003c/li\u003e\n \u003cli\u003eMonteiro JM, Rao A, Shawe-Taylor J, Mour\u0026atilde;o-Miranda J (2016): A multiple hold-out framework for Sparse Partial Least Squares. \u003cem\u003eJ Neurosci Methods\u003c/em\u003e 271: 182\u0026ndash;194.\u003c/li\u003e\n \u003cli\u003ePopovic D, Ruef A, Dwyer DB, Antonucci LA, Eder J, Sanfelici R, \u003cem\u003eet al.\u003c/em\u003e (2020): Traces of Trauma: A Multivariate Pattern Analysis of Childhood Trauma, Brain Structure, and Clinical Phenotypes. \u003cem\u003eBiol Psychiatry\u003c/em\u003e 88: 829\u0026ndash;842.\u003c/li\u003e\n \u003cli\u003eDe Hert M, Dekker JM, Wood D, Kahl KG, Holt RIG, M\u0026ouml;ller H-J (2009): Cardiovascular disease and diabetes in people with severe mental illness position statement from the European Psychiatric Association (EPA), supported by the European Association for the Study of Diabetes (EASD) and the European Society of Cardiology (ESC). \u003cem\u003eEur Psychiatry J Assoc Eur Psychiatr\u003c/em\u003e 24: 412\u0026ndash;424.\u003c/li\u003e\n \u003cli\u003eSv\u0026auml;rd A, Lahti J, M\u0026auml;nty M, Roos E, Rahkonen O, Lahelma E, Lallukka T (2020): Weight change among normal weight, overweight and obese employees and subsequent diagnosis-specific sickness absence: A register-linked follow-up study. \u003cem\u003eScand J Public Health\u003c/em\u003e 48: 155\u0026ndash;163.\u003c/li\u003e\n \u003cli\u003eEder J, Glocker C, Barton B, Sarisik E, Popovic D, L\u0026auml;mmermann J, \u003cem\u003eet al.\u003c/em\u003e (2024): Who is at risk for weight gain after weight-gain associated treatment with antipsychotics, antidepressants, and mood stabilizers: A machine learning approach. \u003cem\u003eActa Psychiatr Scand\u003c/em\u003e. https://doi.org/10.1111/acps.13684\u003c/li\u003e\n \u003cli\u003eMcWhinney SR, Brosch K, Calhoun VD, Crespo-Facorro B, Crossley NA, Dannlowski U, \u003cem\u003eet al.\u003c/em\u003e (2022): Obesity and brain structure in schizophrenia - ENIGMA study in 3021 individuals. \u003cem\u003eMol Psychiatry\u003c/em\u003e 27: 3731\u0026ndash;3737.\u003c/li\u003e\n \u003cli\u003eBaothman OA, Zamzami MA, Taher I, Abubaker J, Abu-Farha M (2016): The role of Gut Microbiota in the development of obesity and Diabetes. \u003cem\u003eLipids Health Dis\u003c/em\u003e 15: 108.\u003c/li\u003e\n \u003cli\u003eWu H, Liu Y, Wang J, Chen S, Xie L, Wu X (2023): Schizophrenia and obesity: May the gut microbiota serve as a link for the pathogenesis? \u003cem\u003eiMeta\u003c/em\u003e 2: e99.\u003c/li\u003e\n \u003cli\u003eBohon C, Welch H (2021): Quadratic relations of BMI with depression and brain volume in children: Analysis of data from the ABCD Study. \u003cem\u003eJ Psychiatr Res\u003c/em\u003e 136: 421\u0026ndash;427.\u003c/li\u003e\n \u003cli\u003ePanariello F, De Luca V, de Bartolomeis A (2011): Weight Gain, Schizophrenia and Antipsychotics: New Findings from Animal Model and Pharmacogenomic Studies. \u003cem\u003eSchizophr Res Treat\u003c/em\u003e 2011: 459284.\u003c/li\u003e\n \u003cli\u003eFava M (2000): Weight gain and antidepressants. \u003cem\u003eJ Clin Psychiatry\u003c/em\u003e 61 Suppl 11: 37\u0026ndash;41.\u003c/li\u003e\n \u003cli\u003eFerns G (2018): Cause, consequence or coincidence: The relationship between psychiatric disease and metabolic syndrome. \u003cem\u003eTransl Metab Syndr Res\u003c/em\u003e 1: 23\u0026ndash;38.\u003c/li\u003e\n \u003cli\u003ePenninx BWJH, Lange SMM (2018): Metabolic syndrome in psychiatric patients: overview, mechanisms, and implications. \u003cem\u003eDialogues Clin Neurosci\u003c/em\u003e 20: 63\u0026ndash;73.\u003c/li\u003e\n \u003cli\u003eAvery JA, Powell JN, Breslin FJ, Lepping RJ, Martin LE, Patrician TM, \u003cem\u003eet al.\u003c/em\u003e (2017): Obesity is associated with altered mid-insula functional connectivity to limbic regions underlying appetitive responses to foods. \u003cem\u003eJ Psychopharmacol Oxf Engl\u003c/em\u003e 31: 1475\u0026ndash;1484.\u003c/li\u003e\n \u003cli\u003eDugr\u0026eacute; JR, Bitar N, Dumais A, Potvin S (2019): Limbic Hyperactivity in Response to Emotionally Neutral Stimuli in Schizophrenia: A Neuroimaging Meta-Analysis of the Hypervigilant Mind. \u003cem\u003eAm J Psychiatry\u003c/em\u003e 176: 1021\u0026ndash;1029.\u003c/li\u003e\n \u003cli\u003eTamminga CA, Thaker GK, Buchanan R, Kirkpatrick B, Alphs LD, Chase TN, Carpenter WT (1992): Limbic System Abnormalities Identified in Schizophrenia Using Positron Emission Tomography With Fluorodeoxyglucose and Neocortical Alterations With Deficit Syndrome. \u003cem\u003eArch Gen Psychiatry\u003c/em\u003e 49: 522\u0026ndash;530.\u003c/li\u003e\n \u003cli\u003eMj H, Ak T, J B, M B, B W (2019): Grey matter alterations in obesity: A meta-analysis of whole-brain studies. \u003cem\u003eObes Rev Off J Int Assoc Study Obes\u003c/em\u003e 20. https://doi.org/10.1111/obr.12799\u003c/li\u003e\n \u003cli\u003eThomas Yeo BT, Krienen FM, Sepulcre J, Sabuncu MR, Lashkari D, Hollinshead M, \u003cem\u003eet al.\u003c/em\u003e (2011): The organization of the human cerebral cortex estimated by intrinsic functional connectivity. \u003cem\u003eJ Neurophysiol\u003c/em\u003e 106: 1125\u0026ndash;1165.\u003c/li\u003e\n \u003cli\u003eBaker JT, Holmes AJ, Masters GA, Yeo BTT, Krienen F, Buckner RL, \u0026Ouml;ng\u0026uuml;r D (2014): Disruption of Cortical Association Networks in Schizophrenia and Psychotic Bipolar Disorder. \u003cem\u003eJAMA Psychiatry\u003c/em\u003e 71: 109.\u003cstrong\u003e\u003c/strong\u003e\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"BMI, machine learning, grey matter volume, schizophrenia, major depressive disorder","lastPublishedDoi":"10.21203/rs.3.rs-5259910/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5259910/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eUnderstanding the neurobiological underpinnings of weight gain could reduce excess mortality and improve long-term trajectories of psychiatric disorders. We used support-vector machines and whole-brain voxel-wise grey matter volume to generate and validate a BMI predictor in healthy individuals (N\u0026thinsp;=\u0026thinsp;1504) and applied it to individuals with schizophrenia (SCZ,N\u0026thinsp;=\u0026thinsp;146), clinical high-risk states for psychosis (CHR,N\u0026thinsp;=\u0026thinsp;213) and recent-onset depression (ROD,N\u0026thinsp;=\u0026thinsp;200). We computed BMIgap (BMI\u003csub\u003epredicted\u003c/sub\u003e-BMI\u003csub\u003emeasured\u003c/sub\u003e), interrogated its brain-level overlaps with SCZ and explored whether BMIgap predicted weight gain at 1- and 2-year follow-up. SCZ (BMIgap\u0026thinsp;=\u0026thinsp;1.05kg/m\u003csup\u003e2\u003c/sup\u003e) and CHR individuals (BMIgap\u0026thinsp;=\u0026thinsp;0.51 kg/m\u003csup\u003e2\u003c/sup\u003e) showed increased and ROD individuals (BMIgap=-0.82 kg/m\u003csup\u003e2\u003c/sup\u003e) decreased BMIgap. Shared brain patterns of BMI and SCZ were linked to illness duration, disease onset, and hospitalization frequency. Higher BMIgap predicted future weight gain, particularly in younger ROD individuals, and at 2-year follow-up. Therefore, we propose BMIgap as a potential brain-derived measure to stratify at-risk individuals and deliver tailored interventions for better metabolic risk control.\u003c/p\u003e","manuscriptTitle":"BMIgap: a new tool to quantify transdiagnostic brain signatures of current and future weight","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-12-11 13:30:42","doi":"10.21203/rs.3.rs-5259910/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"nature-mental-health","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"natmentalhealth","sideBox":"Learn more about [Nature Mental Health](https://www.nature.com/natmentalhealth/)","snPcode":"44220","submissionUrl":"https://mts-natmentalhealth.nature.com/cgi-bin/main.plex","title":"Nature Mental Health","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature Research","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"b5d92ab8-9fd8-466e-9ab3-28582c212725","owner":[],"postedDate":"December 11th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":41429253,"name":"Health sciences/Medical research/Outcomes research"},{"id":41429254,"name":"Health sciences/Diseases/Psychiatric disorders/Psychosis"},{"id":41429255,"name":"Health sciences/Diseases/Psychiatric disorders/Depression"},{"id":41429256,"name":"Health sciences/Diseases/Psychiatric disorders/Schizophrenia"},{"id":41429257,"name":"Health sciences/Biomarkers/Predictive markers"}],"tags":[],"updatedAt":"2025-10-21T07:08:13+00:00","versionOfRecord":{"articleIdentity":"rs-5259910","link":"https://doi.org/10.1038/s44220-025-00522-3","journal":{"identity":"nature-mental-health","isVorOnly":false,"title":"Nature Mental Health"},"publishedOn":"2025-10-20 04:00:00","publishedOnDateReadable":"October 20th, 2025"},"versionCreatedAt":"2024-12-11 13:30:42","video":"","vorDoi":"10.1038/s44220-025-00522-3","vorDoiUrl":"https://doi.org/10.1038/s44220-025-00522-3","workflowStages":[]},"version":"v1","identity":"rs-5259910","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5259910","identity":"rs-5259910","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}