Abstract
Background
Brain connectivity abnormalities may underlie affective psychosis which may be influenced by sex. Using diffusion magnetic resonance imaging (dMRI) tractography, we assessed frontostriatal brain wiring connectivity patterns in 46 early psychosis affective and 56 healthy control young adult, male and female participants.
Methods
We collected harmonized whole brain tractography dMRI data from three sites and applied our novel fiber clustering methodology to identify 17 white matter fiber clusters connecting the frontal cortex and caudate per hemisphere in each group. By comparing cortical and caudate inter-cluster endpoint distances between groups, we assessed their patterns of frontostriatal connectivity.
Results
1) In both males and females in early affective psychosis and healthy controls, bilaterally, the frontal cortex to caudate wiring pattern deviated from a topographic organization. 2) in a right hemisphere frontal pole fiber cluster, independent of sex, we found a significant interaction effect of diagnosis on the connectivity pattern between frontal cortex and caudate, (p = 0.013); 3) in a left hemisphere inferior frontal gyrus, pars triangularis, fiber cluster, we found sex influenced the significant interaction effect of diagnosis on the connectivity pattern between frontal cortex and caudate, (p = 0.0085), with the effect much greater in females (p < 0.00001) than in males (p = 0.97); and, 4) post-hoc tests showed a significant effect of diagnosis on the pattern of frontal cortex to caudate connectivity in specific subregions (ps ≤0.04).
Conclusions
Importantly, while the frontostriatal wiring pattern in both early affective psychosis and healthy controls deviated from a topographic organization, we found in a right hemisphere fiber cluster, independent of sex, that a significant interaction effect of diagnosis was observed on the frontal cortex to caudate connectivity pattern, whereas, in a left hemisphere fiber cluster, the effect of diagnosis on connectivity pattern was influenced by sex, showing the importance of the role of sex in brain wiring patterns in early affective psychosis.
Introduction
Affective psychosis is characterized by a combination of mood symptoms and psychotic thinking. As brain wiring is established during early development [1], the presence of an altered pattern in young adult subjects would support an understanding of affective psychosis as possibly developmental in nature, similar to our understanding in non-affective psychosis [2,3,4,5,6].
Previously, we compared non-affective young adult, early psychosis subjects with the same healthy control subjects used in the present study. In the non-affective study, we found the overall wiring pattern between the frontal cortex and caudate for both groups did not conform to a strictly topographic relationship, i.e., adjacent areas in the cortex did not strictly remain spatially preserved in their projection zones in the caudate. However, we also found regionally specific group wiring pattern differences, namely, we found a diagnostic effect on the frontal cortex to caudate connectivity pattern coming from rostral middle frontal gyrus, frontal pole, and inferior frontal gyrus [2].
We chose the period of early psychosis to mitigate the confounding effects of medication and chronicity. Further, we assessed male and female affective psychosis subjects to identify any influence of sex on the brain wiring patterns which may underlie early affective psychosis. Structural brain dysconnectivity, or brain miswiring, is thought to characterize non-affective psychosis e.g., [2, 7, 8]. Whether such brain miswiring is also present and characterizes affective psychosis in a similar manner is not known and is the principal question to be addressed in this study.
Moreover, distinct measures of brain miswiring in early affective subjects, compared with healthy controls, could serve as developmental biological trait markers. Such biological trait measures would allow for the early identification of individuals who might develop either affective or non-affective psychosis and, thus, afford a more precise early therapeutic intervention. Lastly, as our subjects included both males and females, this allowed us to consider the role of sex in brain wiring in affective psychosis. This is likely relevant as there are many clinical and neurobiological features that differ between the sexes in psychosis and in affective disorders and it is thus important to establish whether neurobiological markers of illness differ between the sexes [9,10,11].
To better understand the focus here on frontostriatal connectivity, it is important to appreciate the anatomy of the basal ganglia circuitry. The main input nucleus of the basal ganglia for projections from the cortex is the striatum which sends feedback through the output nuclei of the basal ganglia, via the thalamus, back to the cortex [12, 13]. Corticostriatal connectivity has been divided into three topographically organized functional subloops. These subloops, in turn, define the three functional subregions of the striatum, i.e., limbic, associative and sensorimotor [14]. Cortical projections to the striatum from the cortex are thought to have a generally topographic connectivity pattern, that is, adjacent areas in the cortex are spatially preserved in their projection zones in the striatum. Such projections also are thought to have an anatomically parallel, functionally segregated, anatomic organization [12]. However, non-topographic (e.g., anatomically converging), functionally integrative projections from the cortex to the striatum also overlap at the level of the striatum [13, 15]. Such integration has been shown both in animal tract tracing and human imaging studies [13, 15,16,17]. It has been suggested that striatal target zones that receive non-topographic, overlapping cortical projections from separate cortical functional subregions, particularly limbic and associative cortex, permit the integration of information beyond simply relaying information [13, 18]. The demonstration of deviations from normal connectivity patterns, particularly in the associative subloop, would likely interfere with cognitive functions subserved by this subloop The primary focus of this paper is on the caudate nucleus as this nucleus is a central component of the associative subloop which subserves brain functions such as executive control and emotion regulation, and goal directed behavior, thought to be disrupted in non-affective and affective psychosis [19,20,21,22,23,24].
In our prior studies, using our current fiber clustering methodology [25], we assessed healthy control (HC) subjects to establish a normal pattern of frontoststriatal brain wiring pattern so that we could contrast this with other neuropsychiatric conditions. We found that the normative prefrontal cortex to caudate brain wiring pattern deviated from a strictly topographic, parallel, organization, due to regionally specific fiber clusters emanating from ventrolateral, dorsolateral and orbitofrontal prefrontal cortex. Although such a general pattern also characterized the early non-affective subjects, in addition, we showed differences in the wiring pattern between HCs and non-affective early psychosis in specific clusters coming from the right hemisphere frontal pole, rostral middle frontal gyrus and inferior frontal gyrus [2].
In the current study, as in our prior study of non-affective psychosis [2], we assessed the frontostriatal brain wring connectivity pattern in subjects with early affective psychosis compared to that of HC subjects. We, again, employed our novel approach to the use of tractography employing a fiber clustering method [26]. Using machine learning, this approach generates clusters of streamlines, also called fibers, based on their trajectory similarity. Using this approach, we identified and assessed the degree of geometric convergence for all fiber clusters, in pairs, connecting regions of interest (ROIs) in the frontal cortex and the caudate. To do this, we measured the mean inter-cluster distance between the endpoints of streamlines of fiber clusters at the level of frontal cortex (i.e., cortical distance) and the mean endpoint distance of the corresponding streamlines of fiber clusters at the level of the caudate (i.e., caudate distance; see Fig. 1). We note that the clusters coming from the frontal cortex that we selected for study have the dorsal striatum, and more specifically the caudate nucleus, as their target zone. Using the caudate as our target zone, we believe, is sufficient for our purpose as it is a major component of the associative striatum, a region that is primarily targeted by the associative cortex (e.g., [27, 28]). The associative striatum is involved in processing prefrontal cortex associative cortical inputs and, hence, is involved in cognitive control and emotion regulation, core deficits in affective psychosis [23, 24]. For technical reasons, described in our Methods below, we omitted assessing the putamen, although part of it is included in the associative striatum, to avoid mismeasurement of convergence between cluster pairs [27, 28].
A model of our method for obtaining cortical and caudate mean distances between the endpoints of streamlines of two frontostriatal fiber clusters. (Adapted from Levitt et al., Cereb Cortex 2021; 31(12): 5308-5318, Oxford University Press.).
To summarize, we here explore whether frontostriatal brain wiring is similar or different in early affective psychosis subjects and HCs. In this study, we assess frontostriatal brain wiring in affective early psychosis young adults with either a bipolar disorder with psychotic features (n = 42) or a major depression with psychotic features (n = 4). Based on our prior study in non-affective early psychosis and healthy subjects [2] and the phenomenological and brain differences in males and females in affective disorders [10, 11, 29], we hypothesized 1) frontostriatal connectivity in both groups would show an overall pattern that would deviate from a topographic arrangement, implying that integration of information, not just relaying of information, occurs subcortically in the striatum; 2) frontostriatal connectivity between early affective psychosis and HCs will show localized differences not observed in our earlier study ; and, 3) the role of sex will affect both overall and localized group differences in frontostriatal wiring pattern.
Methods
Subjects
This study included data from 46 young adult EP-AFF subjects, 21 male, 25 female, and 56 HCs, 37 male, 19 female, from the Human Connectome Project for Early Psychosis (HCP-EP) (MPI: Shenton, Breier [30]). The subjects were recruited from four HCP-EP sites (Indiana University, Massachusetts General Hospital, McLean Hospital and Beth Israel Deaconess Medical Center-Massachusetts Mental Health Center). MRI data were acquired from three HCP-EP sites, one in Indiana, one at Brigham and Women’s Hospital (with subjects from McLean, Massachusetts General Hospital, and Beth Israel Deaconess Medical Center-Massachusetts Mental Health Center) and one later in the study at McLean Hospital.
Ethics approval and consent to participate
Ethical approval for the study was obtained from the Partners Institutional Review Board (BIDMC, McLean, and MGH; IRB00012706) and the Indiana University Institutional Review Board (IRB00000222), and all procedures were performed in accordance with the regulations set forth by the Declaration of Helsinki. All subjects who participated in the study provided informed consent. The brain imaging data and behavioral data are available publicly in data from the HCP-EP Project which is stored at The National Institute of Mental Health Data Archive (NDA).
All early psychosis participants had onset of their illness within 5 years of entry into the study. Affective subjects consisted of individuals diagnosed with bipolar I with psychotic features, other specified bipolar disorder with psychotic features, bipolar II with psychotic features and major depressive disorder with psychotic features (See Table 1). While it is not fully known whether early psychosis major depression and early psychosis bipolar disorder represent distinct neurobiological subtypes, as the goal in this study was to compare affective psychosis subjects with healthy controls, these clinical subtypes by design were combined. HC subjects did not have a history of psychotic disorders or other Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-V) disorders including mood disorder with psychosis; all participants were excluded with known IQ less than 70 or a recent history of substance current within 900 days history of severe substance use disorder. (See Jacobs et al. [30] for a full listing of subject inclusion and exclusion criteria in the HCP-EP Project.)
We note that the sample in this study is smaller than the eventual sample collected from the HCP-EP with the last 2 years of collection occurring during the COVID-19 pandemic. As described in Jacobs et al. [30], for the final study sample, 281 participants had diffusion magnetic resonance imaging (dMRI) data (93% of the 303 participants in the study). Because of COVID, our data analyses were initiated prior to the completion of the HCP-EP project, with a sample resulting in 210 participants who had dMRI data (75% of the 281 participants).
Image acquisition and preprocessing
We acquired the HCP-EP dMRI data using a modified HCP-Lifespan protocol with addition of 6 extra low b-value gradient directions at b = 500 (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6649668/) [2, 30]. dMRI scans were collected on a 3 T Siemens Prisma scanner at all sites. Image processing was performed with the use of our well-established pipeline (https://github.com/pnlbwh) that included motion correction, eddy current correction and echo planar imaging (EPI) distortion correction. The acquisition parameters were: echo time (TE) = 89.2 ms, repetition time (TR) = 3230 ms, partial fourier of 6/8, multi-band factor of 4 and voxel size = 1.5 × 1.5 × 1.5 mm3. We acquired data along 206 gradients in both anterior to posterior (AP) and posterior to anterior (PA) phase encoding directions at b-values of b = 500, 1500 and 3000 s/mm^2 with 6 gradient directions at b = 500 s/mm2 and 100 gradients along each of b = 1500 and 3000 s/mm2.
Structural image post-processing
Diffusion imaging data harmonization
As previously described in [2, 30], we harmonized the dMRI data employing our published harmonization methodology [31] (https://github.com/pnlbwh/dMRIharmonization). To account for site differences, control subjects from each site were used to learn cross-site mapping, with BWH serving as the reference site. These learned mappings were then applied to harmonize dMRI data from EP-AFF subjects across sites to the BWH reference. As shown in our earlier works, our harmonization algorithm preserves statistical group differences despite nonlinear relationships between the imaging and demographic variables [31,32,33].
Diffusion imaging two-tensor whole brain tractography post-processing
We computed whole brain tractography in the harmonized dMRI data using a two-tensor model Unscented Kalman Filter (UKF) method [34, 35], which accounts for crossing fibers, as implemented in the ukftractography package (https://github.com/pnlbwh/ukftractography). Of note, we have shown that the UKF method has very high reproducibility [26, 36]. The first tensor in the two-tensor model represents the main direction of the fiber tract, whereas the second tensor represents fibers that cross through the tract. Visual and quantitative quality control of the tractography was performed using the quality control tool in the whitematteranalysis (WMA) software (https://github.com/SlicerDMRI/whitematteranalysis). This method can be used with open-source code and is available at https://github.com/pnlbwh/ukftractography along with 3D Slicer software (http://www.slicer.org). We extracted the b = 3000 shell of 92 gradient directions and collected 18 b = 0 images, 9 in the AP and 9 in the PA direction interspersed during the scan, in scans for each subject, as applied in our previous studies that perform tractography-based analysis using HCP data [26, 36,37,38]. Angular resolution is better and more accurate at high b-values such as 3000 [39, 40] and thus we used this single shell for performing tractography.
Specifically, all voxels were seeded for tractography within the brain mask where FA was greater than 0.1. For deterministic tractography, an approach that seeds tractography in the entire brain and then selects tracts of interest has been shown to be superior to seeding from specific ROIs [41,42,43,44]. Tracking was stopped when the FA value fell below 0.08 or the normalized mean signal (the sum of the normalized signal across all gradient directions which was used to differentiate between white/gray matter and cerebrospinal fluid) fell below 0.06, as recommended in [26]. Only streamlines exceeding 40 mm were retained to avoid bias towards less reliable short fibers [26, 45]. For each subject, about 700,000 fibers were present in the whole brain tractography.
Fiber clustering methods
Cluster analysis
As previously described in [2, 25], we used a data-driven fiber clustering atlas to identify fiber clusters from orbital, lateral and medial prefrontal cortical regions that project to the caudate [26]. Using this atlas, the results of the brain tractography were parcellated into 2000 fiber clusters according to the geometry of the fiber projections in cerebral white matter. The clusters themselves were comprised of fibers, also called streamlines, reflecting the trajectory of white matter axons (See Fig. 1).
Briefly, we generated the atlas by creating tractography maps of 100 HCP subjects (an independent sample from the HCP subjects used here) and then applied our fiber clustering method to group the tracts across subjects according to their similarity in shape and location. Our atlas provides for each cluster in it, a tract anatomical profile (TAP) [26], i.e. the set of segmented brain FreeSurfer regions through which the cluster passed. We calculated the TAP based on the 100 HCP-atlas subjects, as described in [26]. In brief, for each cluster, the set of intersected FreeSurfer regions (here, a given ROI in the prefrontal cortex and the caudate) per atlas subject was computed. Then, we selected for inclusion in the study from this set of regions those clusters in each hemisphere which were intersected by at least 40% of fiber streamlines of this cluster across all atlas subjects. This resulted in 17 fiber clusters of interest, which we, in turn, used to generate streamlines of interest (see Figure S1).
We used the following FreeSurfer cortical ROIs [46]: the caudalanteriorcingulate, caudalmiddlefrontal, lateralorbitofrontal, medialorbitofrontal, parsopercularis, parsorbitalis, parstriangularis, rostralanteriorcingulate, rostralmiddlefrontal, superiorfrontal, frontalpole, and the caudate nucleus. We targeted the caudate, but not the putamen, in the dorsal striatum. We did this because our inter-cluster endpoint distance measures are Euclidian distances and inter-cluster streamlines that project both to the putamen and the caudate might be measured as close to each other when, assessed using non-Euclidian distances, they are not. To avoid this potential problem, we employed the more simple banana shape of the caudate, alone, for the target of projecting frontal cortex streamlines.
Inter-cluster streamline endpoint distance analysis
As previously described in [2, 25]. to quantify the topographical relationship of fiber clusters, we calculated inter-cluster endpoint distances. For each of the fiber cluster pairs, we measured the mean Euclidean distance between the endpoints of streamlines at the level of the frontal cortex and in the corresponding fiber cluster pair at the level of the caudate (see Fig. 1). This permitted us to quantify, using a Convergence quotient (CQ) described below under Statistical Analysis, the degree of convergent or divergent deviation from a strictly parallel, topographic organization among the 17 frontostriatal fiber cluster projections from the frontal cortex to the caudate. We note, our method does not quantify fiber cluster shape, itself.
Statistical analysis
Fiber clusters, and inter-cluster streamline endpoint distances were determined from the HCP-EP Project diffusion scans on 102 affective subjects and 56 controls. Analyses were performed on data on all available subjects, at the time (See under Subjects above). For all statistical tests, we reported p-values, test statistics, and degrees of freedom. We measured central tendency using means, and error bars are deviations of plus or minus one standard error or of ½ standard deviation.
To determine the pattern of frontostriatal connectivity within both groups, in both sexes, we generated scatter plots for each hemisphere (see Fig. 2a-d) based on the 17 fiber clusters (with 136 pairs of fiber clusters, yielding 136 data points), that showed the relationship between the cortical distances and the corresponding caudate distances of the fiber cluster pairs connecting the prefrontal cortex and the caudate. Here, we analyzed sex separately to avoid the potential confound of differences in head size between sexes. We then fit an exponential model describing the connectivity pattern was then fit to the data points as previously shown in Levitt et al. [25].
Scatter plots of female EP-AFFs and HCs which show the relationship between frontal cortex and caudate inter-cluster endpoint distances in (a) LH EP-AFFs, (b) LH HCs, (c) RH EP-AFFs and (d) RH HCs. Note that the EP-AFF female convex curve in the LH (a) is more flattened in comparison to the other curves. Also note, in Fig. 2, the 16 green circles in each subplot represent clusters that include Cluster 6 as one of its cluster pairs. LH left hemisphere, RH right hemisphere, HCs healthy controls, EP-AFF early psychosis affective subjects; (male plots available on request).
Between group data analysis consisted of between-group comparisons of distances between 17 fiber tracts. Each tract is a cluster of streamlines with cortical and sub-cortical endpoints. The spatial arrangement of the 17 tracts in an individual is, of course, complex. We simplified the problem by assessing the topology by means of the convergence quotient (CQ) for each of the 136 pairs of fiber clusters, per hemisphere, in each subject. The convergence quotient was defined as CQ = (Cortex Distance - Caudate Distance)/ (Cortex Distance + Caudate Distance), for each cluster pair in each cluster, where Cortex Distance and Caudate Distance are the distances between two clusters at the cortical and caudate endpoints, respectively. A cluster pair with endpoints farther apart in cortex than caudate (CQ > 0) is said to be convergent; for the opposite situation, CQ < 0, the pair is divergent. The boundary situation, CQ = 0, corresponds to “parallel” clusters.
Analyzing all 136 cluster pairs in a single statistical model is problematic as such a model would result in a large number of predictors relative to the sample size. We addressed this issue by fitting 17 statistical models for each hemisphere, using mixed-model regression analysis for CQ in each of these 34 models.
For each hemisphere, models are indexed by a reference cluster, taken to be each of clusters 1–17, in turn. Only CQ values based on cortical and sub-cortical distances in a hemisphere between this reference cluster and each of the other 16 clusters were used. Hence, a fixed predictor pair with 16 levels was included in each regression. Sex and Diagnosis were also included as fixed effects, and subject ID was modeled as a random effect. The between-group difference in CQ was thus assessed using the diagnosis by pair interaction. The dependence of the quantity on sex was assessed by testing for significance of the 3-factor interaction involving sex, diagnosis and pair.
We first tested the three-factor interaction, Bonferroni adjusting p-values for the 17 models fit in each hemisphere. If the 3-factor interaction for a given model was not significant, then we tested for a 2-factor diagnosis by pair interaction. We did this by using stepwise regression, requiring that the model selected by the stepwise procedure included the Pair by Diagnosis interaction, and that this interaction was significant, Bonferroni-adjusted as above.
We then explored the nature of these significant diagnosis by pair interactions in more detail using post-hoc hypothesis tests on each coefficient for cluster pair in the clusters that showed a group difference. For each of these two post-hoc analyses, we adjusted for multiple comparisons using the method of Hothorn et al. [47].
To assess missing data affecting Cluster R2 (see the Results section), we tested the association of baseline demographic and neuropsychological characteristics with missing cluster 2, in both affective and control groups (Table S2). The only significant association (based on chi-square tests) was for Sex: in the affective (p = 0.035), but not the control (p = 0.21) group (Tables S1 & S2). Thus, we investigated the association of sex, diagnosis, and their interaction with missing cluster 2 using logistic regression models, with missingness as response. We found that only sex was a significant predictor but not diagnosis or sex by diagnosis. To demonstrate the robustness of our R2 results to missing data, we believe, it suffices to analyze R2 data separately for males and females, combine the p-values using the Fisher’s method commonly used for this purpose [48] in order to arrive at a consensus, and, lastly, compare the results of the three models considered: males only, females only and all subjects. To do this last step, we obtained the average between group difference in fitted values for each of the 16 cluster pairs with Cluster R2, for the full model of our first submission, as well as for models fit to males and females alone and compared them for agreement, using pairwise Pearson correlations.
Results
Subject demographics
102 young adult subjects were included in this study from the Human Connectome-EP project including 46 EP-AFF subjects (45.7% males; 54.3% females) and 56 HCs (66:1% males; 33.9% females) χ2(1) = 04.29, p = 0.038). Affective psychosis subjects’ mean age was 24.02 ± 4.37; HC subjects mean age was 23.84 ± 3.99 (t = (1100)0.22; p = 0.83). We performed a two-way ANOVA with sex and diagnosis and their interaction as independent variables and age as a dependent variable. This showed that there were no significant effects on age of either sex (F = 0.641, 98; p = 0.36), diagnosis (F = 0.231, 98; p = 0.88), or their interaction (F = 1.1181, 98; p = 0.29). These analyses support that the age of the subjects in our study was not associated with sex, diagnosis, or their interaction.
For Full scale IQ, affective psychosis subjects (n = 43; 3 subjects had missing data) had a mean score = 110.3 ± 0.13.9; HC subjects (n = 55; 1 subject had missing datum) had a mean score = 115.4 ± 11.0 (t = (1,96)-2.04; p = 0.044). The vocabulary T score, reflecting premorbid IQ, was not significant between groups (p = 0.266). The mean duration of illness for early psychosis affective subjects was = 565.5 + 523.8 days. See Table 1. For further information including cognitive measures, duration of illness, chlorpromazine (CPZ)-equivalents, educational level, clinical measures, and diagnostic subtypes.
The overall pattern of frontostriatal connectivity, including all cluster pairs, in each sex, within both groups
To evaluate the overall pattern of connectivity between the frontal cortex and striatum, in both groups in both hemispheres, we generated scatter plots in each hemisphere, which show the within group correlation between the cortical cluster endpoint distances at the level of the cortex and corresponding cluster endpoint distances at the level of the caudate for each of the 136 fiber cluster pairs (see Fig. 2a-d).
As we previously found in our HCs and early psychosis-nonaffective (EP-NA) groups [2], we here report a similar finding of a non-linear frontostriatal connectivity pattern in both HCs and affective psychosis groups, but in this study, we analyzed each sex separately in both the left and right hemispheres. As before, for both groups and both sexes, the shapes of the curves, hence their non-linearity, were was driven by the results from the same 10 cluster pairs highlighted by the green circles located in the right lower portion of the scatter plot. See the Supplementary section for the technical details that support this assertion. Of note, in Fig. 2a-d, cluster 6, comes from the inferior frontal gyrus, pars triangularis, which is one of the cluster pairs in each of the 16 green circles. It was significantly over-represented in the 10 cluster pairs driving the non-linearity as revealed by it being a member of each of these 10 cluster pairs for both groups and both sexes and in both hemispheres. Based on combinatorics, there is less than one chance in a billion that this pattern would occur by chance alone (see Fig. 2a-d).
Of further note, we found that the concave correlation curve in the left hemisphere in female early psychosis affective subjects, qualitatively appeared more flattened compared to both male and female controls in both hemispheres and compared to the female early affective psychosis subjects in the right hemisphere, visually suggesting that the wiring pattern in the LH in female early psychosis subjects differed from its pattern in the RH and from the pattern seen in male and female HCs in both hemispheres (see Fig. 2a-d.; male plots available on request).
A comparison between groups of the degree of convergence in frontostriatal connectivity
To compare groups, we used linear mixed model regression analyses. For each of 17 clusters in each hemisphere we assessed the between-group difference in the degree of convergence for the 16 pairs and tested for the dependence of a diagnosis x cluster pair interaction on sex. This revealed 2 regionally localized group differences from 2 separate clusters in the pattern of frontal cortex to caudate brain connectivity, one seen across groups (i.e., a relationship which is the same for sexes) and one seen in females but not in males (i.e., a relationship which is different for both sexes). See Figure S2.
First, we found in cluster 2 coming from RH frontal pole, after not finding an effect of sex, a significant 2-way diagnosis by cluster pair interaction (χ2 = (15, N = 76); = 36.59, adjusted p = 0.013; Bonferroni adjusted, multiplied by 17); see Fig. 3A. & Fig. 4A). Second, we found in cluster 8 coming from LH inferior frontal gyrus, pars triangularis, a 3-way sex x diagnosis x cluster pair interaction (χ2(15, N = 102) = 39.72; adjusted p = 0.0085; Bonferroni adjusted, multiplied by 17), with the diagnosis x cluster pair interaction much greater in females (χ2(15, N = 102) = 69.65; adjusted p < 0.00001) than in males (χ2(15, N = 102) = 6.34; adjusted p = 0.9734) when analyzing each sex separately (see Fig. 3B & Fig. 4B). Also, see Fig. 3C. for a plot of LH cluster 8 showing the significant divergence in pattern for males versus females, ordered by values for females, when comparing subtracted data averaged mean CQs (EP-AFFs -HCs) across 16 cluster pairs. In the above paragraph, all adjusted p-values are adjusted are Bonferroni adjusted, multiplied by 17.
A. Using a mixed model regression analysis of CQs, covarying for sex, we showed a significant group by cluster pair interaction for RH Cluster 2 (p = 0.013096; adjusted). B. Using a mixed model regression analysis of CQs, we showed a significant 3-way sex by diagnosis by cluster pair interaction for LH Cluster 8 (p = 0.008505; adjusted). We note that fiber cluster 2 originates in the frontal pole and fiber cluster 8 originates in the inferior frontal gyrus, pars triangularis. (See Fig. 4A, B). C. A plot of LH cluster 8 showing the significant divergence in pattern for males versus females, ordered by values for females, when comparing subtracted data averaged mean CQs (EP-AFFs -HCs) across 16 cluster pairs. LH left hemisphere. CQ convergence quotient, Aff early psychosis affective subjects, ctrl healthy controls.
A. Superior views of fiber cluster 2 originating in the RH Frontal Pole (4. a.). B. Superior views of fiber cluster fiber cluster 8 originating in Inferior Frontal Gyrus, Pars Triangularis (4. b.). A anterior direction, S superior direction, R right direction. RH right hemisphere.
Third, we explored the nature of these significant diagnosis by pair interactions in more detail using post-hoc hypothesis tests. For the RH frontal pole, analyzing both sexes combined, post-hoc tests (adjusted, using Hothorn et al. [46]) showed a significant effect of diagnosis for 2 pairs of clusters coming from right frontal pole (Clusters R2-16; R2-17) (z = 3.9; adjusted p = 0.00146; z = 2.9; adjusted p = 0.04122). See Fig. 3A. Fourth, and lastly, for cluster L8 coming from LH inferior frontal gyrus, pars triangularis, assessing females alone in a post-hoc analysis, we found a significant diagnosis x cluster pair interaction (z = 3.11; adjusted p = .0374) for a single cluster pair which projected from both the inferior frontal gyrus, pars triangularis and from the rostral middle frontal gyrus (Clusters L8-L3) (see Fig. 3B). In the above paragraph all adjusted p-values are adjusted using Hothorn et al. [47].
Due to scan or anatomical variability, we anticipated finding a small percentage of clusters to be missing in subjects, i.e., not generating streamlines connecting the frontal cortex with the caudate, as we have observed in prior studies [26, 49]. In this study, we found that 31 out of the 102 subjects had missing data for certain clusters. These included 14 HCs (7 females and 7 males) and 17 EAFF subjects (14 females and 3 males) yielding 42 HCs (12 females, 30 males) and 29 EAFF subjects (11 females, 18 males); (χ2 = (1, N = 71);0.69, p = 0.41 for sex proportion).
However, of note, 26 subjects were missing cluster 2 coming from the RH frontal pole but no subjects were missing cluster 8 coming from the LH inferior frontal gyrus. A distinctive aspect of cluster 2 is that the streamlines in it are short, which may have led to the unusual number of subjects missing this cluster due to our 40 mm cut-off criterion for streamline inclusion. See the Discussion section below where we discuss this further. These subjects missing RH cluster 2 included 11 HCs (6 females and 5 males) and 15 EAFF subjects (12 females and 3 males) yielding sample sizes of 45 HCs (32 males and 13 females) and 31 EAFFs subjects (18 males and 13 females); (χ2 = (1, N = 76);1.39, p = 0.24 for sex proportion).
Importantly, our use of a mixed model regression analysis, in contrast to a mixed model ANOVA, makes it possible to conduct analyses where there is missing data. For our above mixed model regression analysis involving cluster 2, all cases missing cluster 2 streamlines connecting the frontal cortex to the caudate were automatically omitted from the analysis, which was thus based on 76 subjects. For the mixed model regression analysis involving cluster 8, we based our analysis on all 102 subjects. Although there were 26 subjects missing the single cluster 8-2, notably, our analysis allowed for preserving the very great majority of the remaining clusters pairs between cluster 8 and the other 15 clusters.
Lastly, as noted above. of 102 subjects, 26 were missing cluster 2 data (15 Affective, 11 Control). We investigated the sensitivity of our results to these missing data by testing the association of missingness with diagnosis, sex, and other baseline characteristics. Only sex in the affective group was significant (P = 0.035). We next investigated the association of sex, diagnosis, and their interaction with missing cluster 2 using logistic regression models, with missingness as response. Sex as a single predictor is significant (p = 0.002), but diagnosis is not significant (p = 0.13), nor is the sum of diagnosis plus the interaction (p = 0.54). We therefore performed separate analyses in males and females. The p-values for the diagnosis by cluster pair interaction is 0.03 for males (N = 50), 0.13 for females (N = 26), which yields a combined p-value (Fisher’s method [48]) of 0.03. We then compared the diagnosis x pair interactions predicted for the three models considered (all data, males only, females only) using Pearson correlation, obtaining pairwise correlations > 0.9, demonstrating excellent agreement.
Discussion
In this study, we assessed the organization of frontostriatal brain wiring in early psychosis affective subjects and healthy controls using a novel diffusion imaging fiber cluster tractography method. There were five major findings: (1) when we analyzed affective and control groups within groups rather than between groups, we found in both sexes the overall wiring pattern between frontal cortex and caudate deviated from a topographic organization (defined further below). Additionally, when we analyzed between group differences in brain connectivity, there were three additional localized findings. (2) In a right hemisphere frontal pole fiber cluster, independent of sex, we found a significant interaction effect of diagnosis on the pattern of connectivity between frontal cortex and caudate. (3) In a left hemisphere inferior frontal gyrus, pars triangularis, fiber cluster, we found sex influenced the interaction effect of diagnosis on the pattern of connectivity pattern between frontal cortex and caudate. (4) Analyzing both sexes combined, post-hoc tests showed a significant effect of diagnosis for two pairs of clusters in the RH coming from the frontal pole and from the rostral middle frontal gyrus. (5) Finally, assessing females alone, post-hoc tests showed a significant effect of diagnosis for a single cluster pair in the LH which projected from both the inferior frontal gyrus, pars triangularis and from the rostral middle frontal gyrus. These significant post-hoc cluster pair findings suggest that these cluster pairs are especially relevant for driving the pattern of difference in connectivity in the above described two clusters where we showed a significant effect of diagnosis on the pattern of connectivity. This implies also that prefrontal cortical subregions from which these cluster pairs originate have different frontal cortex to caudate connectivity patterns between diagnostic groups. The potential clinical relevance of this is discussed below.
Our finding of an overall bilateral pattern of connectivity between the frontal cortex and caudate, for all subjects, that deviated from a strictly topographic arrangement is consistent with our prior studies in non-affective early psychosis and in healthy controls [2, 25]. Namely, adjacent areas in the cortex are not strictly spatially preserved in their projection zones in the caudate, which allows for the integration, instead of simply the relay, of information at the level of the caudate. That is, inputs from different functional subregions of the frontal cortex can converge in the caudate. However, we also showed both overlapping and different localized differences between the two subtypes of early psychosis (affective and non-affective) and heathy controls. In our prior study of non-affective early psychosis [2], combining sexes, we showed a difference in wiring pattern for two right hemisphere clusters projecting from the frontal pole, rostral middle frontal gyrus and inferior frontal gyrus, pars orbitalis. Whereas here we demonstrate that combining sexes shows a difference in wiring pattern in a right hemisphere frontal pole cluster and in females, alone, a difference in wiring pattern in a left hemisphere inferior frontal gyrus, pars triangularis cluster. Overall, this supports both a transdiagnostically shared wiring pattern abnormality in frontostriatal circuitry for both sexes across early psychosis subtypes in a region in the right hemisphere prefrontal cortex, as well as a diagnostically specific difference between early psychosis subtypes in females, alone, in a region in the left hemisphere prefrontal cortex. A further approach we are pursuing is to cluster all early psychosis subjects, including both subtypes, based on cognitive measures, and then to compare group differences across these data-based derived cognitive clusters with group differences based on DSM-V diagnostic categories. We believe this approach will enable us to further embed our current brain wiring findings in a transdiagnostic or Research Domain Criteria framework.
Our left hemisphere findings emphasize the importance of considering sex in early psychosis and raise the possibility of regional sex differences in brain structures that might help to explain sex differences in the clinical presentation in early affective psychosis. Such sex differences in demographic, clinical and neuropsychological features in bipolar disorder have been reported. For example, a recent literature search of large sample (>1000 subjects) studies of bipolar disorder patients published after 2010 showed more females patients in both bipolar 1 and bipolar II [50], demonstrating demographic sex differences in bipolar disorder. Further, Menculini et al. [10] found that women had more severe symptoms of mania while men had more mixed features demonstrating clinical sex differences in bipolar disorder. Sole et al., [11] reported neuropsychological sex differences, albeit small, where females outperformed males in verbal learning, and males outperformed females in working memory [11]. Lastly, male sex was found to be a risk factor for poorer social cognition bipolar disorder subjects (Varro, 2022) [51], but this was not supported in another study [52].
Regarding features of brain structure, previous studies have similarly demonstrated sex differences in bipolar disorder. For example, studies assessing brain volume and sex differences in normal subjects report increased gray matter, white matter and CSF volumes in males vs females, e.g., [53] as well as multiple selective cortical gray matter regions showing sexual dimorphisms [9]. For example, one study by Luders et al. [54], reported a number of regional differences between the sexes in normal young adults in gray matter concentration with greater gray matter concentration in females in the left hemisphere inferior frontal gyrus, pars triangularis, which is where the current study shows a sex difference in wiring pattern in early affective psychosis. Interestingly, in a study performing graph analysis using dMRI diffusion tensor imaging tractography [55], these investigators found one of multiple regions distinguished females from males which included the left hemisphere inferior frontal gyrus, pars triangularis, the structure where we here show a sex effect on brain wiring between affective psychosis subjects and healthy controls.
A strength of our approach, which assesses the macrostructural pattern of brain wiring, is that that it could provide a trait biomarker for early psychosis. As brain wiring occurs early in development and is stable after birth, its measurement in young adulthood could serve as a trait biomarker for disorders such as schizophrenia with abnormal early development [3, 6, 56], or in affective psychosis with possible early abnormal development. Of interest, tract tracing studies in monkeys have shown a distinct pattern of corticostriatal connectivity that is established in the final third of pregnancy prior to birth [57], and importantly, this pattern persists in adult monkeys [58]. Genes and prenatal stress that influence neuronal migration and axonal growth, potentially altering brain connectivity of white matter long tracts, similarly could cause brain miswiring in affective psychosis as well as in schizophrenia [59,60,61].
The cortico-basal ganglia circuitry regulates cortical function by providing both positive and negative feedback to concurrently activated cortical areas projecting to the basal ganglia, hence influencing higher cognitive functions as well as motor function [14, 62, 63]. As our data here show group differences in clusters originating in the right hemisphere frontal pole independent of sex and in the left hemisphere inferior frontal gyrus, pars triangularis dependent on sex, it is of interest to review further the function of these prefrontal cortex subregions.
The frontal pole is described as involved in cognitive functions including multi-tasking, prospective memory, and mentalizing (i.e., theory of mind), functions which have been subsumed under the term metacognition [19, 64,65,66,67,68,69,70,71,72]. Further, it has been proposed that the lateral frontal pole enables humans to perform “branching operations”, i.e., the ability to multitask by suppressing a prior goal directed behavior kept in working memory while pursuing another goal, allowing for enhanced cognitive flexibility in humans [73].
Deficits in the aforementioned functions are also reported in schizophrenia. However, a metanalysis comparing cognition in first-episode bipolar disorder patients with first-episode schizophrenia patients and healthy controls show qualitatively similar deficits, though quantitively less severe, in first-episode bipolar compared with first-episode schizophrenia patients [74]. Another function proposed for the medial frontal pole, part of the default mode network, is to inhibit negative affect in depression [75, 76]. As bipolar disorder subjects over the course of illness are more frequently depressed than manic e.g., [77, 78], miswiring of connectivity emanating from the frontal pole could be relevant for bipolar disorder, and bipolar depression, in particular. In this study, we propose that frontal cortex to caudate miswiring emanating from specific prefrontal cortical subregions that subserve specific cognitive functions may alter such functions in early psychosis.
Also relevant to functions of the frontal lobe, the inferior frontal gyrus (i.e., the ventrolateral prefrontal cortex), which consists of Brodmann Area (BA) 44, 45, and 47, has been associated with higher cognitive functions, such as “spatial attention, inhibitory control, and language” [65] and is activated during word retrieval and working memory updating [79]. Further, Broca’s area (44/45) of the left inferior frontal gyrus is thought to be involved in a large left hemisphere language network influencing phonological and semantic processes [80, 81]. Of importance for our study, the inferior frontal gyrus, pars triangularis is Brodman area 45.
An additional role for the left inferior gyrus is in social cognition [82, 83]. For example, Diveia et al. [83] report that the anterior left inferior frontal gyrus region participates both in semantic control and in social cognition, suggesting “that social cognition is partly regulated by the neurocognitive system underpinning semantic control.” Our data support the possibility that the organization of connectivity for language processing and social cognition differs between female and male bipolar subjects. As noted above, given that women outperformed men in the verbal learning and memory recognition tasks, regardless of being patients or healthy controls [11], it is of interest that underlying brain wiring in an area which subserves language processing and social cognition is the area in which we found a selective sex group difference in brain wiring pattern. Moreover, some propose that the left hemisphere inferior frontal gyrus subserves impulsivity, and is over-activated in bipolar subjects during reward anticipation and is also a potential biomarker for mania/hypomania in at-risk bipolar disorder subjects [23, 84,85,86].
Of further interest for our findings is that the cluster pair where we found a significant diagnosis x cluster pair interaction, in a post hoc analysis in females, alone, project from the inferior frontal gyrus, pars triangularis (i.e., ventrolateral prefrontal cortex) and rostral middle frontal gyrus (i.e., dorsolateral prefrontal cortex), respectively. Tan et al., [87] report that a disrupted balance in activation between dorsolateral and ventrolateral prefrontal cortex, predominantly in the left hemisphere, is associated with executive dysfunction in first-episode schizophrenia which, based on our findings, might similarly operate in female bipolar psychosis. Also, combining sexes, post-hoc testing showed group differences in the wiring pattern of two clusters between the right hemisphere frontal pole and the right hemisphere rostral middle frontal gyrus that connect the default mode network and the fronto-parietal network. Evidence supports that the default mode network, active during mind wandering, is suppressed when shifting to tasks requiring active attention to external stimuli [88] Thus, miswiring between the frontal pole, part of the default mode network, and the caudate, part of the frontoparietal network, which subserves active attention to external stimuli, might worsen this cognitively important network interaction.
There are nonetheless several limitations in the current study. The reliability and utility of our CQ-based measures of brain wiring in future studies should be assessed through replication studies that assess affective psychosis subjects across the lifespan including in clinically at high-risk subjects, in subjects with chronic illness and in studies with a longitudinal design. The demonstration of the stability and reliability of our measures would strengthen the idea that our brain wiring measures represent trait biomarkers. Other limitations include the confounding effects of medication and illness chronicity that cannot be fully ruled out as many of the patients were taking psychotropic agents, including antipsychotics, and the durations of illness varied. An additional limitation, as described in the Methods section, is that individual streamlines within fiber clusters for our study need not terminate directly onto the caudate to qualify as streamlines counted in the endpoint calculations. A further limitation is that some subjects were missing cluster data. However, we note that our use of a mixed model regression analysis allows for missing data. Additionally, as described in the Results and Methods sections, despite the missing data, we are able to demonstrate the robustness of our finding for Cluster R2. Further, due to scan or anatomical variability, we expected to find a small percentage of clusters to be missing in subjects as has been the case in prior studies of ours [26, 49]. Here we found that there were 31 out of 102 subjects who had missing clusters. As described above, cluster 2 streamlines coming from the right hemisphere, frontal pole, were missing in 26 subjects. We believe this occurred because this cluster was comprised of short fibers, and as such, short fiber connections in smaller brains may not pass the length threshold of 40 mm for including streamlines that we used. Before creating the atlas, we removed streamlines under 40 mm from each atlas subject’s whole brain tractography data to avoid bias towards less reliable short fibers [26, 45]. In our next atlas, we plan to have a more lenient length threshold. Lastly, using dMRI tractography can result in both false negative and false positive streamlines [89, 90].
In summary, using our novel dMRI tractography method to assess the geometry of brain wiring organization in early affective psychosis, showed that the overall pattern of connectivity between the frontal cortex and caudate was not strictly topographic, similar to what we found in our previous study of non-affective early psychosis. However, we also showed a different pattern of brain wiring in the frontal cortex to caudate circuitry between affective and non-affective early psychosis subjects compared with healthy controls, and in this study of affective early psychosis, where we considered the effect of sex, we showed its importance in brain wiring patterns. The above suggests diagnostic specificity for the wring patterns of early psychosis affective and non-affective subtypes. As brain wiring occurs early in development and is stable after birth, its measurement in young adulthood could serve as a trait biomarker for early psychosis disorders in a diagnostically specific manner and, thus, might guide more precise early treatment interventions.
Code availability
We performed statistical analyses using R (www.r-project.org), including the “lme4” and “lmerTest” libraries for mixed model regression. The R code scripts employed will be made available upon request. We computed whole brain white matter tractography using the unscented Kalman filter (UKF) tractography (https://github.com/pnlbwh/ukftractography). We performed Tractography fiber clustering using the whitematteranalysis (WMA) package (https://github.com/SlicerDMRI/whitematteranalysis) and the O’Donnell Research Group (ORG) atlas (http://dmri.slicer.org/atlases/), provided via the SlicerDMRI software (http://dmri.slicer.org). Code for inter-cluster endpoint distance analysis will be made available upon request.
+References
Kandel E, Schwartz J, Jessell T. Principles of neural science. New York: McGraw-Hill, Health Professions Division; 2000.
Levitt JJ, Zhang F, Vangel M, Nestor PG, Rathi Y, Cetin-Karayumak S, et al. The organization of frontostriatal brain wiring in non-affective early psychosis compared with healthy subjects using a novel diffusion imaging fiber cluster analysis. Mol Psychiatry. 2023;28:2301–11.
Murray RM, Lewis SW. Is schizophrenia a neurodevelopmental disorder? Br Med J. 1987;295:681–2.
Ripke S, Sanders AR, Kendler KS, Levinson DF, Sklar P, Holmans PA, et al. Genome-wide association study identifies five new schizophrenia loci. Nat Genet. 2011;43:969–76.
Sekar A, Bialas AR, de Rivera H, Davis A, Hammond TR, Kamitaki N, et al. Schizophrenia risk from complex variation of complement component 4. Nature. 2016;530:177–83.
Weinberger DR. Implications of normal brain development for the pathogenesis of schizophrenia. Arch Gen Psychiatry. 1987;44:660–9.
Levitt JJ, Nestor PG, Kubicki M, Lyall AE, Zhang F, Riklin-Raviv T, et al. Miswiring of frontostriatal projections in schizophrenia. Schizophr Bull. 2020;46:990–8.
Stephan KE, Baldeweg T, Friston KJ. Synaptic plasticity and dysconnection in schizophrenia. Biol Psychiatry. 2006;59:929–39.
Goldstein JM, Seidman LJ, Horton NJ, Makris N, Kennedy DN, Caviness VS Jr., et al. Normal sexual dimorphism of the adult human brain assessed by in vivo magnetic resonance imaging. Cereb Cortex. 2001;11:490–7.
Menculini G, Steardo L Jr., Sciarma T, D’Angelo M, Lanza L, Cinesi G, et al. Sex differences in bipolar disorders: impact on psychopathological features and treatment response. Front Psychiatry. 2022;13:926594.
Sole B, Varo C, Torrent C, Montejo L, Jimenez E, Bonnin CDM, et al. Sex differences in neurocognitive and psychosocial functioning in bipolar disorder. J Affect Disord. 2022;296:208–15.
Alexander GE, Crutcher MD. Functional architecture of basal ganglia circuits: neural substrates of parallel processing. Trends Neurosci. 1990;13:266–71.
Haber SN. The primate basal ganglia: parallel and integrative networks. J Chem Neuroanat. 2003;26:317–30.
Redgrave P, Rodriguez M, Smith Y, Rodriguez-Oroz MC, Lehericy S, Bergman H, et al. Goal-directed and habitual control in the basal ganglia: implications for Parkinson’s disease. Nat Rev. 2010;11:760–72.
Averbeck BB, Lehman J, Jacobson M, Haber SN. Estimates of projection overlap and zones of convergence within frontal-striatal circuits. J Neurosci. 2014;34:9497–505.
Draganski B, Kherif F, Kloppel S, Cook PA, Alexander DC, Parker GJ, et al. Evidence for segregated and integrative connectivity patterns in the human Basal Ganglia. J Neurosci. 2008;28:7143–52.
Lehericy S, Ducros M, Van de Moortele PF, Francois C, Thivard L, Poupon C, et al. Diffusion tensor fiber tracking shows distinct corticostriatal circuits in humans. Ann Neurol. 2004;55:522–9.
Haber SN. Convergence of limbic, cognitive, and motor cortico-striatal circuits with dopamine pathways in primate brain. In: Iversen LL, Iversen SD, Dunnett SB, Bjorklund A, eds. Dopamine handbook. Oxford: Oxford University Press, Inc.; 2010. pp. 38–48.
Barch DM, Dowd EC. Goal representations and motivational drive in schizophrenia: the role of prefrontal-striatal interactions. Schizophr Bull. 2010;36:919–34.
Howes OD, Montgomery AJ, Asselin MC, Murray RM, Valli I, Tabraham P, et al. Elevated striatal dopamine function linked to prodromal signs of schizophrenia. Arch Gen Psychiatry. 2009;66:13–20.
Kegeles LS, Abi-Dargham A, Frankle WG, Gil R, Cooper TB, Slifstein M, et al. Increased synaptic dopamine function in associative regions of the striatum in schizophrenia. Arch Gen Psychiatry. 2010;67:231–9.
Levitt JJ, Nestor PG, Levin L, Pelavin P, Lin P, Kubicki M, et al. Reduced structural connectivity in frontostriatal white matter tracts in the associative loop in schizophrenia. Am J Psychiatry. 2017;174:1102–11.
Phillips ML. Building a neurobiological framework to elucidate neural mechanisms, aid early risk detection, and develop new treatments for bipolar disorder. Psychiatry Res. 2023;329:115521.
Sabaroedin K, Tiego J, Fornito A. Circuit-based approaches to understanding corticostriatothalamic dysfunction across the psychosis continuum. Biol Psychiatry. 2023;93:113–24.
Levitt JJ, Zhang F, Vangel M, Nestor PG, Rathi Y, Kubicki M, et al. The organization of frontostriatal brain wiring in healthy subjects using a novel diffusion imaging fiber cluster analysis. Cereb Cortex. 2021;31:5308–18.
Zhang F, Wu Y, Norton I, Rigolo L, Rathi Y, Makris N, et al. An anatomically curated fiber clustering white matter atlas for consistent white matter tract parcellation across the lifespan. NeuroImage. 2018;179:429–47.
Haber SN. Corticostriatal circuitry. Dialogues Clin Neurosci. 2016;18:7–21.
Levitt JJ, Rosow LK, Nestor PG, Pelavin PE, Swisher TM, McCarley RW, et al. A volumetric MRI study of limbic, associative and sensorimotor striatal subregions in schizophrenia. Schizophr Res. 2013;145:11–19.
Ziemka-Nalecz M, Pawelec P, Ziabska K, Zalewska T. Sex differences in brain disorders. Int J Mol Sci. 2023;24:14571.
Jacobs GR, Coleman MJ, Lewandowski KE, Pasternak O, Cetin-Karayumak S, Mesholam-Gately RI, et al. An introduction to the human connectome project for early psychosis. Schizophr Bull. 2025;51:658–71.
Cetin Karayumak S, Bouix S, Ning L, James A, Crow T, Shenton M, et al. Retrospective harmonization of multi-site diffusion MRI data acquired with different acquisition parameters. NeuroImage. 2019;184:180–200.
Cetin-Karayumak S, Di Biase MA, Chunga N, Reid B, Somes N, Lyall AE, et al. White matter abnormalities across the lifespan of schizophrenia: a harmonized multi-site diffusion MRI study. Mol Psychiatry. 2020;25:3208–19.
Mirzaalian H, Ning L, Savadjiev P, Pasternak O, Bouix S, Michailovich O, et al. Inter-site and inter-scanner diffusion MRI data harmonization. NeuroImage. 2016;135:311–23.
Malcolm JG, Shenton ME, Rathi Y. Filtered multitensor tractography. IEEE Trans Med Imaging. 2010;29:1664–75.
Reddy CP, Rathi Y. Joint multi-fiber NODDI parameter estimation and tractography using the unscented information filter. Front Neurosci. 2016;10:166.
Zhang F, Wu Y, Norton I, Rathi Y, Golby AJ, O’Donnell LJ. Test-retest reproducibility of white matter parcellation using diffusion MRI tractography fiber clustering. Hum Brain Mapp. 2019;40:3041–57.
O’Donnell LJ, Suter Y, Rigolo L, Kahali P, Zhang F, Norton I, et al. Automated white matter fiber tract identification in patients with brain tumors. Neuroimage Clin. 2017;13:138–53.
Zhang F, Norton I, Cai W, Song Y, Wells WM, O’Donnell LJ. Comparison between two white matter segmentation strategies: an investigation into white matter segmentation consistency. IEEE 14th international symposium on biomedical imaging (ISBI 2017). 2017. pp 796-9.
Descoteaux M, Angelino E, Fitzgibbons S, Deriche R. Regularized, fast, and robust analytical Q-ball imaging. Magn Reson Med. 2007;58:497–510.
Ning L, Westin CF, Rathi Y. Estimating diffusion propagator and its moments using directional radial basis functions. IEEE Trans Med Imaging. 2015;34:2058–78.
Essayed WI, Zhang F, Unadkat P, Cosgrove GR, Golby AJ, O’Donnell LJ. White matter tractography for neurosurgical planning: a topography-based review of the current state of the art. Neuroimage Clin. 2017;15:659–72.
Huang H, Zhang J, van Zijl PC, Mori S. Analysis of noise effects on DTI-based tractography using the brute-force and multi-ROI approach. Magn Reson Med. 2004;52:559–65.
Radmanesh A, Zamani AA, Whalen S, Tie Y, Suarez RO, Golby AJ. Comparison of seeding methods for visualization of the corticospinal tracts using single tensor tractography. Clin Neurol Neurosurg. 2015;129:44–49.
Wakana S, Caprihan A, Panzenboeck MM, Fallon JH, Perry M, Gollub RL, et al. Reproducibility of quantitative tractography methods applied to cerebral white matter. NeuroImage. 2007;36:630–44.
Bajada CJ, Schreiber J, Caspers S. Fiber length profiling: a novel approach to structural brain organization. NeuroImage. 2019;186:164–73.
Desikan RS, Segonne F, Fischl B, Quinn BT, Dickerson BC, Blacker D, et al. An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest. NeuroImage. 2006;31:968–80.
Hothorn T, Bretz F, Westfall P. Simultaneous inference in general parametric models. Biom J. 2008;50:346–63.
Mosteller F. Questions and answers. Am Statistician. 1948;2:30–31.
Cetin-Karayumak S, Zhang F, Zurrin R, Billah T, Zekelman L, Makris N, et al. Harmonized diffusion MRI data and white matter measures from the adolescent brain cognitive development study. Sci Data. 2024;11:249.
Dell’Osso B, Cafaro R, Ketter TA. Has bipolar disorder become a predominantly female gender related condition? Analysis of recently published large sample studies. Int J Bipolar Disord. 2021;9:3.
Varo C, Sole B, Jimenez E, Bonnin CM, Torrent C, Valls E, et al. Identifying social cognition subgroups in euthymic patients with bipolar disorder: a cluster analytical approach. Psychol Med. 2022;52:159–68.
Navarra-Ventura G, Vicent-Gil M, Serra-Blasco M, Massons C, Crosas JM, Cobo J, et al. Group and sex differences in social cognition in bipolar disorder, schizophrenia/schizoaffective disorder and healthy people. Compr Psychiatry. 2021;109:152258.
Good CD, Johnsrude I, Ashburner J, Henson RN, Friston KJ, Frackowiak RS. Cerebral asymmetry and the effects of sex and handedness on brain structure: a voxel-based morphometric analysis of 465 normal adult human brains. NeuroImage. 2001;14:685–700.
Luders E, Narr KL, Thompson PM, Woods RP, Rex DE, Jancke L, et al. Mapping cortical gray matter in the young adult brain: effects of gender. NeuroImage. 2005;26:493–501.
Sun Y, Lee R, Chen Y, Collinson S, Thakor N, Bezerianos A, et al. Progressive gender differences of structural brain networks in healthy adults: a longitudinal, diffusion tensor imaging study. PLoS One. 2015;10:e0118857.
Weinberger DR. Future of days past: neurodevelopment and schizophrenia. Schizophr Bull. 2017;43:1164–8.
Goldman-Rakic PS. The corticostriatal fiber system in the rhesus monkey: organization and development. Prog Brain Res. 1983;58:405–18.
Goldman-Rakic PS. Prenatal formation of cortical input and development of cytoarchitectonic compartments in the neostriatum of the rhesus monkey. J Neurosci. 1981;1:721–35.
Chen SY, Huang PH, Cheng HJ. Disrupted-in-Schizophrenia 1-mediated axon guidance involves TRIO-RAC-PAK small GTPase pathway signaling. Proc Natl Acad Sci USA. 2011;108:5861–6.
Mukai J, Tamura M, Fenelon K, Rosen AM, Spellman TJ, Kang R, et al. Molecular substrates of altered axonal growth and brain connectivity in a mouse model of schizophrenia. Neuron. 2015;86:680–95.
Solek CM, Farooqi NAI, Brake N, Kesner P, Schohl A, Antel JP, et al. Early inflammation dysregulates neuronal circuit formation in vivo via upregulation of IL-1beta. J Neurosci. 2021;41:6353–66.
Liddle PF. Disordered mind and brain: the neural basis of mental symptoms. The Royal College of Psychiatrists; London, 2001.
Girault JA. Integrating neurotransmission in striatal medium spiny neurons. Adv Exp Med Biol. 2012;970:407–29.
Burgess PW, Wu HC. Rostral prefrontal cortex (Brodmann Area 10) metacognition in the brain. In: Stuss DTKRT, ed. Principles of frontal lobe function. 2nd edition. New York: Oxford University Press; 2013. p. 524–44.
Szczepanski SM, Knight RT. Insights into human behavior from lesions to the prefrontal cortex. Neuron. 2014;83:1002–18.
Butler PD, Hoptman MJ, Smith DV, Ermel JA, Calderone DJ, Lee SH, et al. Grant report on social reward learning in schizophrenia (dagger). J Psychiatr Brain Sci. 2020;5:e200004.
Green MF. What are the functional consequences of neurocognitive deficits in schizophrenia? Am J Psychiatry. 1996;153:321–30.
Kring AM, Gur RE, Blanchard JJ, Horan WP, Reise SP. The clinical assessment interview for negative symptoms (CAINS): final development and validation. Am J Psychiatry. 2013;170:165–72.
Oliver LD, Hawco C, Homan P, Lee J, Green MF, Gold JM, et al. Social cognitive networks and social cognitive performance across individuals with schizophrenia spectrum disorders and healthy control participants. Biol Psychiatry Cogn Neurosci Neuroimaging. 2021;6:1202–14.
Palaniyappan L, Simmonite M, White TP, Liddle EB, Liddle PF. Neural primacy of the salience processing system in schizophrenia. Neuron. 2013;79:814–28.
Quan M, Lee SH, Kubicki M, Kikinis Z, Rathi Y, Seidman LJ, et al. White matter tract abnormalities between rostral middle frontal gyrus, inferior frontal gyrus and striatum in first-episode schizophrenia. Schizophr Res. 2013;145:1–10.
Strauss GP, Waltz JA, Gold JM. A review of reward processing and motivational impairment in schizophrenia. Schizophr Bull. 2014;40:S107–116.
Koechlin E, Hyafil A. Anterior prefrontal function and the limits of human decision-making. Science. 2007;318:594–8.
Bora E, Pantelis C. Meta-analysis of cognitive impairment in first-episode bipolar disorder: comparison with first-episode schizophrenia and healthy controls. Schizophr Bull. 2015;41:1095–104.
Alagapan S, Choi KS, Heisig S, Riva-Posse P, Crowell A, Tiruvadi V, et al. Cingulate dynamics track depression recovery with deep brain stimulation. Nature. 2023;622:130–8.
Mayberg HS. Limbic-cortical dysregulation: a proposed model of depression. J Neuropsychiatry Clin Neurosci. 1997;9:471–81.
Judd LL, Akiskal HS, Schettler PJ, Endicott J, Maser J, Solomon DA, et al. The long-term natural history of the weekly symptomatic status of bipolar I disorder. Arch Gen Psychiatry. 2002;59:530–7.
Tondo L, Vazquez GH, Baldessarini RJ. Depression and mania in bipolar disorder. Curr Neuropharmacol. 2017;15:353–8.
Borst JP, Anderson JR. Using model-based functional MRI to locate working memory updates and declarative memory retrievals in the fronto-parietal network. Proc Natl Acad Sci USA. 2013;110:1628–33.
Costafreda SG, Fu CH, Lee L, Everitt B, Brammer MJ, David AS. A systematic review and quantitative appraisal of fMRI studies of verbal fluency: role of the left inferior frontal gyrus. Hum Brain Mapp. 2006;27:799–810.
Ries SK, Xie K, Haaland KY, Dronkers NF, Knight RT. Role of the lateral prefrontal cortex in speech monitoring. Front Hum Neurosci. 2013;7:703.
Binney RJ, Ramsey R. Social semantics: the role of conceptual knowledge and cognitive control in a neurobiological model of the social brain. Neurosci Biobehav Rev. 2020;112:28–38.
Diveica V, Koldewyn K, Binney RJ. Establishing a role of the semantic control network in social cognitive processing: a meta-analysis of functional neuroimaging studies. NeuroImage. 2021;245:118702.
Arora M, Chase H, Bertocci MA, Skeba AS, Eckstrand K, Bebko G, et al. Left ventrolateral prefrontal cortical activity during reward expectancy and mania risk. JAMA Psychiatry. 2025;82:274–84.
Bertocci MA, Chase HW, Graur S, Stiffler R, Edmiston EK, Coffman BA, et al. The impact of targeted cathodal transcranial direct current stimulation on reward circuitry and affect in bipolar disorder. Mol Psychiatry. 2021;26:4137–45.
Caseras X, Lawrence NS, Murphy K, Wise RG, Phillips ML. Ventral striatum activity in response to reward: differences between bipolar I and II disorders. Am J Psychiatry. 2013;170:533–41.
Tan HY, Choo WC, Fones CS, Chee MW. fMRI study of maintenance and manipulation processes within working memory in first-episode schizophrenia. Am J Psychiatry. 2005;162:1849–58.
Buckner RL, DiNicola LM. The brain’s default network: updated anatomy, physiology and evolving insights. Nat Rev. 2019;20:593–608.
Maier-Hein KH, Neher PF, Houde JC, Cote MA, Garyfallidis E, Zhong J, et al. The challenge of mapping the human connectome based on diffusion tractography. Nat Commun. 2017;8:1349.
Thomas C, Ye FQ, Irfanoglu MO, Modi P, Saleem KS, Leopold DA, et al. Anatomical accuracy of brain connections derived from diffusion MRI tractography is inherently limited. Proc Natl Acad Sci USA. 2014;111:16574–9.
Marder SR, Davis JM, Chouinard G. The effects of risperidone on the five dimensions of schizophrenia derived by factor analysis: combined results of the North American trials. J Clin Psychiatry. 1997;58:538–46.
Funding
R21MH121704 (Levitt, Nestor, Zhang, Rathi, Shenton); U01MH104977 (Shenton, Breier, Ongur, Holt, Keshavan; Bouix, Kubicki, Pasternak, Lewandowski, Vangel, Coleman); R01MH119222 (Rathi, O’Donnell, Zhang); R01MH125860 (O’Donnell, Rathi, Zhang); NARSAD Young Investigator Award (Cetin-Karayumak), K24 MH110807 and R01MH112748 (Kubicki), VA Merit Award (I01 CX000176-06; Shenton).
Author information
Authors and Affiliations
Contributions
JJL: conceptualization and design of the study, methodology, data analysis, writing original draft, manuscript review and editing. FZ: conceptualization and design of the study, methodology, software, data analysis, manuscript review and editing. MV: conceptualization and design of the study, methodology, software, statistical analysis, manuscript review and editing. PGN: provided input on science, review of manuscript. YR: conceptualization, methodology, manuscript review and editing. SC-K: provided input on science, methodology, review of manuscript. MK: provided input on science, manuscript review. MJC: data science, QA/QC of clinical database, review of manuscript. OP: design and execution of MRI data processing and quality control, review of manuscript. KEL: provided input on science, data collection, review of manuscript. DH: provided input on science, data collection, review of manuscript. MKesh: provided input on science, data collection, manuscript review and editing. SB: design and execution of MRI data processing and quality control, review of manuscript. DO: provided input on science, data collection, review of manuscript. AB: provided input on science, data collection, review of manuscript. MES: provided input on science, oversaw collection of data, as well as QA/QC and data processing, manuscript review and editing. LJO: conceptualization and design of the study, methodology, software, manuscript review and editing.
Corresponding author
Ethics declarations
Competing interests
+The authors declare no competing interests.
+ +Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Levitt, J.J., Zhang, F., Vangel, M. et al. Diagnostic and sex effects on frontostriatal brain wiring structural organization in early psychosis affective subjects compared to healthy controls. + Mol Psychiatry 31, 6498–6509 (2026). https://doi.org/10.1038/s41380-026-03719-9
Published:
Version of record:
Issue date:
DOI: https://doi.org/10.1038/s41380-026-03719-9
Subjects
+-
+
+
- + Diagnostic markers + +
- + Health sciences + +
- + White Matter Connectivity in Social Cognition + +



