Content uploaded by Audrey Louise Berardi
Author content
All content in this area was uploaded by Audrey Louise Berardi on Mar 14, 2025
Content may be subject to copyright.
Bipolar Disorders, 2025; 0:1–12
https://doi.org/10.1111/bdi.70010
1 of 12
Bipolar Disorders
ORIGINAL ARTICLE OPEN ACCESS
White Matter, Cognition, and Electrophysiological Variables
in Bipolar Disorder: Using Multimodal Integration of
Biomarker Variables Associated With Bipolar Disorder to
Elucidate Deficits
AudreyBerardi1 | JenniferA.Brown1 | BrookeS.Jackson1 | Ling-YuHuang2 | RebekahL.Trotti2 | DavidA.Parker1,3 |
ScotK.Hill4 | ElenaIvleva5 | GodfreyD.Pearlson6 | CarolA.Tamminga5 | MatcheriS.Keshavan7 | SarahK.Keedy8 |
ElliotS.Gershon8 | JohnA.Sweeney9 | BrettA.Clementz1 | JenniferE.McDowell1
1Department of Psychology, University of Georgia, Athens, Georgia, USA | 2Beth Israel Deaconess Medical Center, Boston, Massachusetts,
USA | 3Department of Human Genetics, Emor y School of Medicine, Atlanta, Georgia, US A | 4Department of Psychology, Rosalind Franklin University of
Medicine and Science, Chicago, Illinois, USA | 5Department of Psychiatr y, University of Texas Southwestern Medical Center, Dallas, Texas, USA | 6The
Institute of Living, Hartford, Connecticut, USA | 7Department of Psychiatry, Harvard Medical School, Boston, Massachusetts, USA | 8Department
of Psychiatry and Behavioral Neuroscience, University of Chicago, Chicago, Illinois, USA | 9Department of Psychiatry and Behavioral Neuroscience,
University of Cincinnati, Cincinnati, Ohio,USA
Correspondence: Audrey Berardi (aberardi@uga.edu)
Received: 1 July 2024 | Revised: 3 Februar y 2025 | Accepted: 10 February 2025
Funding: Funding was provided by the National Institute of Mental Health: grant number MH096957 to GDP, MH0969 42 to MSK, MH096913 to CAT, and
MH096900 to BAC. DP and RLT were supported by MH117315, UL1TR0 02378/TL1TR002382 .
Keywords: bipolar disorder| cognition| diff usion tensor imaging (DTI)| electrophysiolog y| intrinsic EEG
ABSTRACT
Aim: This study aimed to evaluate associations in bipolar disorder (BD) across multimodal measures of white matter microstruc-
ture (using diffusion tensor imaging; DTI), cognitive, behavioral, and brain electrophysiological measures (using electroenceph-
alography; EEG).
Methods: Subjects were recruited through the Psychosis and Affective Research Domains and Intermediate Phenotypes
Consortium (n = 45 bipolar with psychosis, n = 40 bipolar without psychosis, n = 66 healthy subjects). DTI data were used to
quantify the white matter variables, fractional anisotropy (FA) and radial diffusivity (RD). The Brief Assessment of Cognition
in Schizophrenia (BACS), Stop Signal Task (SST), pro- and anti- saccades, auditory event- related potentials (ERPs), and intrinsic
brain activity were used as estimates of brain function.
Results: The combined BD group differed from healthy controls, but no differences between BD with and without psychosis
were observed. BD- related white matter abnormalities were seen across multiple tracts: right cingulum– cingulate gyrus, bilateral
anterior thalamic radiation, bilateral superior longitudinal fasciculus, right inferior longitudinal fasciculus, and forceps major.
Results also showed modestly compromised cognitive performance and elevated intrinsic EEG activity associated with BD.
Conclusions: Further analysis indicated worse white matter integrity related to higher intrinsic EEG and modestly higher ERPs.
These multimodal analyses are likely to aid in creating future informative diagnostic, etiological, and treatment targets for BD.
This is a n open access ar ticle under the terms of t he Creative Commons Attr ibution-NonCommercial-NoDer ivs License, whi ch permits use and d istribution in any me dium, provided th e original
work is properl y cited, the use is non- commercial and no mo difications or a daptations are ma de.
© 2025 T he Author(s). Bipolar Disorders published by Joh n Wiley & Sons Ltd.
2 of 12 Bipolar Disorders, 2025
1 | Introduction
The current gold standard for psychiatric diagnosis and research
globally is the Diagnostic and Statistical Manual of Mental
Disorders [1] (5th edition; DSM- 5), which relies on symptoms
and signs for classification. One diagnostic decision of special
interest involves bipolar disorder (BD), a psychiatric condition
that can include psychotic symptoms. In the DSM, the diagnos-
tic decision tree starts with assessing for an affective syndrome
before determining if psychosis is an accompanying feature.
However, persons with BD with psychosis are more likely to
have relatives with a psychotic disorder than those with BD
without psychosis [2]. Persons with schizophrenia have higher
rates of BD with psychosis among their relatives compared to
those with BD without psychosis [2, 3]. These and other results
suggest the initial diagnostic decision should be the presence of
psychosis rather than the presence of an affective syndrome.
Structural brain imaging has been used to probe markers of
psychiatric conditions since the 1980s and has sparked the dis-
covery of profound insights into the nature of mental illness [4].
White matter structural integrity supports interneural commu-
nication, and disrupted white matter relates to the disturbance
of various functions and behaviors [5]. Diffusion tensor imag-
ing (DTI) captures rates of diffusion of water molecules along
axons. White matter integrity can be evaluated using fractional
anisotropy (FA)—a ratio capturing the primary vector of water
movement relative to the orthogonal vectors along the lengths
of axons. The scale goes from zero, meaning isotropic, which
indicates random diffusion, to one, which is anisotropic and is
restricted and follows one direction of flow. Radial diffusivity
(RD) captures the f low orthogonal to the primary vector, captur-
ing flow across fibers, and is inversely correlated with FA. This
study utilized both FA and RD metrics to assess white matter
microstructure in BD.
Two white matter tracts have been found repeatedly to be ir-
regular in BD: the arcuate fasciculus (associated with auditory
processing and speech production) and middle longitudinal
fasciculi (indicated in modulating extra ocular muscle move-
ment s) [6]. The literature is inconsistent, however, regarding
corpus callosum deviations. Yurgelun- Todd et al. [7] reported
in persons with BD an elevation in microstructural differences,
with elevated FA values in the genu of the corpus callosum.
Comparisons of FA values and behavioral symptomology be-
tween schizophrenia probands, BD with psychosis probands,
their first- degree relatives, and healthy persons were also con-
ducted [8]. Skudlarski etal. [8] highlighted a significant clinical
regression effect restricted to the genu of the corpus callosum
for BD with psychosis, such that a continuous decrease in FA
values was noted, with the highest FA values for healthy per-
sons, relatives with no affective clinical features, relatives with
affective clinical features, and bipolar persons [8].
Another useful probe of brain function is electroencephalogra-
phy (EEG). EEG provides insight into neural activation differ-
ences in relation to (a) stimulus presentation, and (b) ongoing
brain activity not specifically tied to stimulus processing (back-
ground, induced, or intrinsic activity). The P300 event- related
potential (ERP) component to oddball stimuli is consistently
reported to be of lower amplitude in BD [9], indicating an
abnormality of context updating in working memory, like that
observed in schizophrenia [10]. The magnitude of this P300 de-
viation may be related to the presence of psychosis in persons
with BD, but there is inconsistency in the literature [11]. It is also
known that not all persons with BD and psychosis share a com-
mon physiolog y [2]. Alternatively, the neural activations in the
first few hundred milliseconds of stimulus processing may dif-
ferentiate BD from other serious psychiatric conditions and BD
with and without a history of psychosis [12]. This may indicate
a fundamental difference in sensory registration among some
persons with BD that may be moderated by familial or constitu-
tional (not necessarily genetic) factors [13]. The relationships of
these deviations to white matter integrity are unknown.
BD is associated with disruptions in cognitive performance that
span alterations in prosody, verbal fluency, and speech produc-
tion [14]. Impairments also have been noted in sustained atten-
tion, memory, and executive functioning [15] among persons
with BD. Clarifying the relationships between cognitive func-
tions and white matter integrity in BD would yield useful in-
formation to assist diagnosis, etiological investigations, and the
development of targeted treatments.
The Bipolar and Schizophrenia Network for Intermediate
Phenotypes (B- SNIP [16]) recruited participants with clinical
diagnoses of psychosis. A related project, the Psychosis and
Affective Research Domains and Intermediate Phenotypes
(PARDIP) study, aimed to understand neurobiological dif-
ferences in BD as a function of psychosis history [13]. Like all
B- SNIP projects investigating the neurobiology of psychosis
[2], participants in PARDIP completed an extensive biomarker
panel. This battery includes measures at multiple levels of
analysis that relate to known or suspected deviations in brain
functioning in persons with serious psychiatric conditions [16].
Measures included cognition, ocular motor (saccadic eye move-
ments), and electrophysiological indices (EEG) of background
(or intrinsic) brain activity, sensory registration, and target
detection- related activations. The joint inclusion of measures
at the cognitive, behavioral, and electrophysiological levels of
analysis is essential for meaningfully probing brain structure–
function relationships. Research efforts have helped emphasize
multimodal relationships between a few of the bio- factor mea-
sures and structural data, including relationships between au-
ditory processing as gauged by EEG activity and microstructure
in the arcuate fasciculus, as well as relationships between ocular
motor movements and microstructure in the middle longitudi-
nal fasciculus [17].
This study aims to elucidate patterns of white matter structural
deviations in persons with BD as a function of the presence of
psychotic features, and how white matter alterations may relate
to cognitive and physiological differences in this population.
To investigate brain structure–function relationships, we used
integrated variables, called bio- factors [18] that span cogni-
tive, behavioral, and electrophysiological levels of analysis. We
hypothesized the following: (a) white matter microstructural
changes would differ across the three groups (persons with bi-
polar with psychosis, persons with bipolar without psychosis,
and healthy persons), (b) cognitive and electrophysiological
biomarkers would differ between groups, particularly in neu-
ral responses to auditory stimuli, and (c) these white matter
13995618, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/bdi.70010 by University Of Georgia Libraries, Wiley Online Library on [14/03/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3 of 12
deviations and neurocognitive alterations would be correlated
with one another. The goal was to create a multimodal integra-
tion of biomarker variables associated with BD.
2 | Methods
2.1 | Participants
Participants were recruited via the multisite PARDIP (Psychosis
and Related Domains Intermediate Phenotypes) consortium,
a B- SNIP subproject, using newspaper and community adver-
tisements. Inclusion and exclusion criteria were the same as
for all other iterations of B- SNIP projects [16]. For the bipolar
groups, diagnoses were made based on the Structural Clinical
Interview for DSM- IV- TR (SCID) [19]. Healthy persons were
screened using the Non- Patient Edition of the SCID [20]. Group
membership and demographics are provided in Table1. In the
sample for analyses, a total of 151 persons (n = 45 bipolar with
psychosis, n = 40 bipolar without psychosis, n = 66 healthy sub-
jects) were recruited across three sites, with ages ranging from
18 to 55 years.
2.2 | Procedures
2.2.1 | DTI Data
For the collection of DTI data, scans were collected on 3 Tesla
magnetic resonance imaging (MRI) systems across 3 sites
(Boston, MA; Dallas, TX; and Hartford, CT, USA). Images were
acquired as described by Brown et al. [21] using a diffusion-
weighted protocol and echoplanar imaging slice sequences.
Sequence parameters were comparable between the different
scanners used (see Table2 for site- specific systems and parame-
ters used during DTI collection).
2.2.2 | Bio- Factor Data
A variety of laboratory measures, as described in Clementz etal.
[2], captured cognitive, behavioral, and electrophysiological per-
formances. Data collection of these laboratory measures was
done using the Brief Assessment of Cognition in Schizophrenia
(BACS) [22], Stop Signal Task (SST) [23], saccadic eye movement
testing with antisaccade and prosaccade tasks [24], intrinsic
neural activity [25], and auditory brain responses to repeated
stimuli (pai red stimulus) as well as ra ndom stimuli (oddball stim-
ulus) [12, 26]. Data across laboratory measures were condensed
to “bio- factors” following the detailed procedures previously de-
scribed [2]. The bio- factors captured: (a) total cognition (BACS),
antisaccade performance (error rates and correct response laten-
cies to gauge visual orienting and goal maintenance); inhibitory
ability under visual conflict (SST); (b) intrinsic neural activity
(intrinsic EEG activity, paired stimulus ongoing high frequency
activity); (c) ERP strength (paired stimulus P1/N1/P2 Complex,
Oddball P1/N1/P2 Complex); and (d) stimulus salience (P300
Complex, Paired Stimulus P2 Response, Latency).
2.3 | Analyses
2.3.1 | DTI Data
Pre- processing steps for DTI were followed in the procedures
outlined in Brown et al. [21]: Raw diffusion images were con-
verted from DICOM (General Electric, Siemens scanners) or
PAR/REC formats (Philips scanners) to NIFTI (nii). A process
of volume removal for motion artifacts was conducted, and vol-
umes with motion were removed from the participant scans and
their respective b- value and b- vector tables. Pre- processing DTI
data was then done using the FMRIB Software Library (FSL);
corrections for eddy current- induced distortions were applied,
and affine transformation was used to register the first non-
weighted image (b = 0). The Brain Extraction Tool was then used
to keep only brain tissue, and a single image for each DTI metric
(FA, RD) was created for each participant using FSL's Diffusion
Toolbox and a tensor was then fit to each white matter voxel. All
FA and RD images were then aligned to MNI152 space using
Tract- Based Spatial Statistics (TBSS) and subsequently con-
verted from nii image files to ascii text files using FSL's fsl2ascii
tool (see Figure1).
These text files were then harmonized using ComBat, a tech-
nique identified in genetic studies to eliminate the influence
of location differences on small- sample- sized datasets, and
TABL E | Participant demographic i nformation for the thre e groups (bipolar with p sychosis, bipolar without ps ychosis, and healthy compar isons).
Number of persons per group, mean age (and standard deviation) in years, sex (% male), ethnicity (% Hispanic), and race (% White, % Black, %
multiracial/other) and mean participant SES score (and standard deviation) are reported.
Var iable
Bipolar with
psychosis (n = 45)
Bipolar without
psychosis (n = 40)
Healthy persons
(n = 66)
Age (years) 40 .6 (± 10. 2) 43.4 (± 11.4) 35.7 (± 12.6)
Sex (% male) 51% 33% 58%
Ethnicity (Hispanic) 13% 10% 9%
Race (White) 67% 73% 61%
Race (Black) 31% 23% 21%
Race (multiracial/other) 2% 4% 18%
Participant SES score 36.6 (± 13.9) 61.4 (± 1 52.6) 32.9 (± 12.7)
13995618, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/bdi.70010 by University Of Georgia Libraries, Wiley Online Library on [14/03/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
4 of 12 Bipolar Disorders, 2025
expanded to DTI analysis [27, 28]. The harmonization was first
carried out with a reference group of only healthy persons and
the resulting values were then applied to the whole sample to
ensure the removal of inter- site variability while preserving the
variance potentially related to psychopathology. For the refer-
ence group, the DTI scalar measurements of FA and RD were
standardized to have similar overall means and variances as
outlined in Johnson et al. [28]. A local and scale adjustment
model was then applied, with the assumption that “…each indi-
vidual covariate is expected to be linearly associated with DTI
scalar measurements” [27]. An empirical Bayes (EB) framework
was used in ComBat by estimating information across metrics
and sites to create better estimates of data points and to stabi-
lize the data set. Estimates generated from the healthy persons
reference group were then used to calculate harmonized values
for the entire sample using the formula (see Figure2).
Harmonized values were then converted back to nii files using
FSL's fslascii2img tool and put back into TBSS and combined
to calculate a mean FA image for the sample, generating an
FA skeleton with the core of common fiber bundles across the
specified sample. FA values were then restricted to voxels con-
taining 18 specific white matter tracts using a combination of
three masks as discussed in Schaeffer etal. [29]. The first mask
applied was used to select voxels with FA greater than 0.2. The
second mask applied, the Johns Hopkins University white mat-
ter tractography atlas, was used to select voxels with a proba-
bility greater than 5% of containing a particular tract. A final
TABLE | Scanner systems at each of the three collection sites and a list of the parameters for each of the sites used to achieve as similar data for
DTI as possible.
Boston Dallas Hartford
Scanners
Magnet—manufacturer GE Philips Siemens
Model HDxt Achieva Allegra
Head coil (channels) 8 8 8
Parameters—diffusion- weighted EPI
TR (m s) 15,000 6300 6300
TE (m s) 85 85 85
Flip angle (°) 90 90 90
FOV (mm) 220 × 22 0 22 0 × 220 1540 × 15 40
Acquisition matrix 128 × 128 128 × 125 128 × 128
Slice thickness (mm) 3 3 3
Voxel size (mm) 1.72 × 1.72 × 3 1.72 × 1.75 × 3 1.72 × 1.72 × 3
Number of slices 52 45 45
Slice orientation Axial Axial Axial
Number of directions 32 32 32
Number of b = 0 images 1 1 1
FIGUR E | Visualization of the masking steps used to remove unwanted white matter f ibers. The first mask removes whole brain FA values
too low (with a threshold of 0.2). The second mask used the Johns Hopkins University White Matter Tractography Atlas, and selected voxels with a
probability greater than 5% of containing a particular tract. The final mask applied to select overlap from voxels from individual scans and the same
voxels in the sample FA skeleton. The combination of these three masks for a particular tract, the forceps major, is shown above.
13995618, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/bdi.70010 by University Of Georgia Libraries, Wiley Online Library on [14/03/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
5 of 12
mask was applied to individual scans to select the same voxels in
the sample FA skeleton. The remaining voxels after the applica-
tion of the three masks were used in analyses. Mean FA and RD
values were calculated for each participant from the remaining
voxels for the 18 specified tracts.
For statistical analyses of DTI variables, an analysis of covari-
ance (ANCOVA) was implemented to investigate differences in
white matter tracts for persons with BD and healthy persons,
with a covariate of sex. Initial analyses were conducted inves-
tigating differences between persons with BD with psychotic
features, BD without psychotic features, and healthy controls.
There were no significant differences for the DTI variables be-
tween bipolar with psychosis and bipolar without psychosis
groups, so they were collapsed for remaining analyses. To ad-
just for possible influences of age, standardized variables were
formed using the healthy persons to model age effects [30].
2.3.2 | Bio- Factor Data
Bio- factor data were quantified using procedures outlined in
Clementz etal. [2, 31]. Each of the resulting factors for the dif-
ferent assessments were then transformed into standardized
units. One- way ANOVAs were conducted to investigate group
differences between individuals with BD and healthy persons
across the different bio- factors. Additionally, analyses of covari-
ance (ANCOVA) were used to investigate the differences in the
bio- factor scores for persons with BD and healthy persons with
a covariate of socioeconomic status, which did not significantly
affect outcomes. The familywise alpha (0.05) in these analyses
was not adjusted for the number of tests because these outcomes
were simply used to screen for variables that may be useful in
the multivariate procedures described below.
2.3.3 | Integrated Analyses
To probe the relationships between DTI outcomes and bio- factor
domains, canonical correlation analyses (CCA) describe the bi-
directional associations between white matter variables (FA and
RD across tracts) and all bio- factor scores. CCA is a data- driven,
multivariate approach that identifies the relationship between
two sets of variables by maximizing correlations between vari-
ables on one side of the equation (white matter integrity) and
variables on the other side of the equation (bio- factors) [32].
This analysis method is the general procedure for investigat-
ing the relationship between two variable sets, and for which
other parametric significance tests are special cases[33]. CCA
creates correlated pairs of latent variates. Each pair is indepen-
dent and composed of weighted sums of the predictor variables
that maximally correlate with the weighted sums of the criterion
variables [33]. Interpretation of what the latent variates repre-
sent is determined by the loadings of individual measures on
the latent structure, much like principal components analysis.
Components of significance were then analyzed via hierarchical
multiple regression analyses, with the first model evaluating the
relationship between white matter and bio- factors, and a second
model evaluating how clinical diagnosis group membership may
affect this relationship.
3 | Results
To evaluate clinical diagnosis group differences, we ran multiple
analyses of covariance (ANCOVAs) for both white matter met-
rics (FA, RD) for the 18 specified white matter tracts. Initially,
ANCOVAs were run with the following group membership:
bipolar with psychosis, bipolar without psychosis, and healthy
persons for both data types (white matter and cognition). No
significant differences were found between the BD groups. To
explore general differences specific to BD compared to healthy
persons, we collapsed the participant groups into a single BD
group. Adjustments for age and sex were included in the statisti-
cal models to investigate and account for their possible influence
on the variables of interest. No such relationships accounted for
the following reported group differences.
Persons with BD had lower FA and higher RD in several tracts
when compared to healthy persons. For tracts that significantly
differed by group, visual representations are shown (red and
yellow; see Figure 3a–e). For those same tracts, mean differ-
ences by group are provided (see Figure4). BD was associated
with significantly reduced FA values in the right cingulum–
cingulate gyrus portion (R- CGC) (F(1,150) = 4.41, p = 0.037)
when compared to healthy values. A complementary result was
found in this same tract (R- CGC) for RD, where persons with
BD exhibited significantly higher RD values (F(1,150) = 9.631,
p = 0.017) compared to healthy values. Persons with BD also
had elevated RD values in bilateral anterior thalamic radiation
FIGUR E | The ComBat harmonization methodology used to calculate the harmonized DTI metrics, FA and RD. The variables are defined as
outlined in Fortin etal. [20], “Let yijv represent the DTI metric (FA, RD) for voxel ν, scan j at site i, αν is the overall DTI metric measure for voxel ν. Χ
is the n × K design matrix for the K covariates of interest (e.g., gender, age), f is a prespecified multivariate function of the covariates outlined by βv.
The multivariate function f is assumed to be the same for all voxels, and that βv is sufficiently capturing all voxel- specific effects. γiv and δiv represent
the additive and multiplicative site effects of site i for voxel v, respectively … f(Χij, βv) is the nonlinear component … [modeled by using] cubic splines.
Cubic splines are composed of piecew ise third- order polynomials with control points (knots) specified in advance. They allow to model nonlinear
relationships bet ween two variables in a flexible and smooth fashion”.
13995618, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/bdi.70010 by University Of Georgia Libraries, Wiley Online Library on [14/03/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
6 of 12 Bipolar Disorders, 2025
(ATR) (left: F(1,150) = 38.052, p < 0.001; right: F(1,150) = 24.378,
p = 0.003), and bilateral superior longitudinal fasciculus (SLF)
(left: F(1,150) = 18.036, p = 0 .023; r ight: F(1,150) = 15.40 3,
p = 0.007) when compared to healthy. Finally, BD participants
had elevated RD values in the right inferior longitudinal fascic-
ulus (ILF) (F(1,150) = 10.340, p < 0.001), and the forceps major
(Fmaj) (F(1,150) = 67.146, p < 0.001) compared to healthy. Effect
sizes as gauged by Glass' Δ are reported in Figure5, exhibiting a
general trend of more disruption for persons with BD compared
to healthy persons.
To investigate the possible associations of socioeconomic status
and age on these measures, a standardized healthy persons' age
variable and participant SES scores from the Hollingshead Two-
Factor Index were included as covariates in ANCOVAs. Use of
the covariates did not influence the statistical outcomes.
Regarding bio- factors, results yielded significant differences on
BACS with all subdimensions (F(1,123) = 24.655, p < 0. 001) an-
tisaccade (F(1,117) = 14.746, p < 0.001), S ST (F(1,121) = 15 .281,
p < 0.001), paired stimulus (PS) ongoing high frequency (HF)
activit y (F(1,139) = 4.319, p = 0.040), and oddball (OB) ongoing
HF activity (F(1,140) = 9.554, p = 0.002) for bipolar persons when
compared to healthy persons (see Table3). A visualization of
mean differences between the two groups (bipolar and healthy
persons) for the bio- factor assessments is included in Figure6.
FIGUR E | (a–e) Visual depictions using Johns Hopkins University W hite Matter Tractography Atlas of significant tracts. Tract s were mapped to
MNI 152 space using FSLeyes in radiological orientation. The color scale shown represents probability of fibers falling into a specific tract, with red
representing an at least 5% likelihood, to yellow representing a 95% likelihood of being part of the specified tract.
Tracts of Significant Group Differences (BP, Healthy Subjects)
a. R Cingulum – Cingulate Gyrus Portion b. Bilateral Anterior Thalamic Radiation
c. Superior Longitudinal Fasciculus d. Right Inferior Longitudinal Fasciculus
e. Forceps Major
13995618, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/bdi.70010 by University Of Georgia Libraries, Wiley Online Library on [14/03/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
7 of 12
Canonical correlation investigated the bidirectional relation-
ships between the two data types—white matter, measured
by DTI metrics FA and RD, and electrophysiological variables
summarized by B- SNIP bio- factors. Two CCA models were
run—one for FA and one for RD. The FA CCA model did not
yield any significant outcomes. The RD CCA model yielded one
significant canonical variate pair (F(1, 81) = 1.234, p = 0.03 4,
r = 0.733). The RD loadings for this component were positively
weighted by the bilateral corticospinal tract (CST) (loading
weight—left: 0.291; loading weight—right: 0.224) and negatively
weighted from other tracts (loading weight—forceps minor:
−0.259; loading weight—right SLF: −0.235). The bio- factor
FIGUR E | DTI metrics (FA, RD) mean and standard error for each tract with significant group differences between persons with bipolar dis-
order (blue) and healthy persons (gray). The complementary results for FA and RD in the right cingulum–cingulate gyrus portion (CGC) illustrate
the level of disruption in this tract in persons with bipolar disorder when compared to healthy persons. Results also indicated elevated RD values
bilaterally in the anterior thalamic radiation (ATR), and bilaterally in superior longitudinal fasciculus (SLF). Elevated RD values were also noted for
the right inferior longitudinal fasciculus (ILF) and forceps major (FMaj). FA and RD results are reported on the same scale, with RD results being
multiplied by 1000.
FIGUR E | Glass' Δ Effect sizes for each of the FA and RD results of significant difference between bipolar and healthy groups. The overall trend
is more deviation in the bipolar disorder group, with Δ values listed for the following tracts: Right cingulum–cingulate gyrus portion (FA = −0.20 0
and R D = −0.271); bilatera l anterior thalamic rad iation RD (left = −0.577, right = −0.542); bilat eral superior longitudinal f asciculus RD (left = −0.264,
right = −0.083); inferior longitudinal fasciculus RD (right = −0.397); forceps major RD (= −0.588) .
13995618, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/bdi.70010 by University Of Georgia Libraries, Wiley Online Library on [14/03/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
8 of 12 Bipolar Disorders, 2025
loadings were primarily weighted by intrinsic activity and ERP
strength (loading weight—Intrinsic EEG Activity (IEA): 0.536;
loading weight—Paired Stimulus P1/N1/P2 Complex: 0.402; see
Table4).
Weights from the first component were then applied to each
individual subject's data, outputting two variables—a vari-
able for the white matter canonical variate, computed based
on the weights of the first component for those variables,
and a variable for the bio- factor variate, computed based on
the weights of the first component for those variables. A plot
of the component- weighted variables from the two sets (see
Figure 7) indicated a positive relationship with overall lower
RD (because of primarily negative loadings), indicative of
worse white matter integrity related to higher intrinsic EEG
activity (more neural noise) and modestly larger ERPs. These
relationships illustrate the possible usefulness of a multi-
modal analysis approach.
4 | Discussion
This project assessed multimodal measures of white matter mi-
crostructure, cognition, behavioral, and brain electrophysiolog-
ical measures in BD with and without psychosis, and healthy
persons. We were interested in white matter differences, as well
as relationships between these measures and bio- factor vari-
ables. Results did not yield significant differences between BD
TABLE | Results of various ANCOVA models investigating group differences between persons with bipolar disorder and healthy persons. Bio-
factor results with significant group differences with covariates of standardized age value based on the healthy persons' group age and participant
socioeconomic score based on the Hollingshead Two- Factor Index.
Bio- factor variable F(df ) pBipolar (n = 8 5) SE Healthy (n = 66) SE
Brief Assessment of
Cognition—Schizophrenia
24.655 (1123) < 0.0 01 −0.002 ± 0 .767 0.712 ± 0.633
Antisaccade 14.746 (1117) < 0.001 0.192 ± 0 .912 −0.464 ± 0 .676
Stop Signal Task 15.281 (1121) < 0.001 −0.208 ± 1.07 0.594 ± 0 .964
Oddball Ongoing High Frequency
Activity
9.554 (1140) 0.002 0.329 ± 0.899 −0.185 ± 0. 891
Paired Stimulus Ongoing High
Frequency Activity
4.319 (1139) 0.04 −0.009 ± 0 .972 −0.390 ± 0.94 4
FIGUR E | Mean standardized unit scores across 11 bio- factor variables for groups assigned by the Diagnostic and Statistical Manual of Mental
Disorders (DSM)- IV diagnoses groups: Bipolar groups (blue) and healthy persons (gray). The variables are grouped to capture the following areas:
(a) cognition (antisaccade, BACS, SST), (b) intrinsic neural activity (intrinsic EEG activity, paired stimulus ongoing high frequency activity), (c) ERP
strength (paired stimulus P1/N1/P2 Complex, Oddball P1/N1/P2 Complex), and (d) stimulus salience (P300 Complex, Paired Stimulus P2 Response,
Latency) .
13995618, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/bdi.70010 by University Of Georgia Libraries, Wiley Online Library on [14/03/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
9 of 12
subgroups, which suggests white matter structure may be less
useful than other laboratory measures for differential diagno-
sis of psychotic features [26]. Differences were found, however,
between the collapsed groups of BD and healthy persons, and
between white matter structure and laboratory measures of
brain function.
TABLE | The results of the RD CCA model for the first component for each of the variables. Loadings correspond with the weight of the variable
on a scale of −1 to 1, with closer to 1 indicating strong positive relationship to the component, and −1 indicating a strong negative relationship to the
component. A heat map was applied to these set loading scores, where −1 is in blue, 0 is in white, and 1 is in red. The left side of the table represents
the largest loa dings for the mean RD v alues for white matter tracts . The right side of the table represents t he loadings f or the bio- f actors and illustrate s
that this component for this set is primarily driven by intrinsic activity and ERP strength related bio- factors.
Var iable — set 1 Component 1 Var iable — set 2 Component 1
Left Corticospinal Tract (RD) 0.291 Intrinsic EEG Activity 0.536
Right Corticospinal Tract (RD) 0.224 Paired Stimulus P1/N1/P2 Complex 0.402
Right Inferior Longitudinal Fasciculus (RD) −0.146 Oddball P1/N1/P2 Complex 0.367
Right Uncinate Fasciculus (RD) −0.156 Brief Assessment of
Cognition—Schizophrenia
0.209
Left Cingulum—Cingulate Gyrus Portion
(R D)
−0.159 Stop Signal Task 0.174
Right Inferior Fronto- Occipital Fasciculus
(R D)
−0.163 Paired Stimulus Ongoing
High Frequency Activity
0.158
Left Inferior Fronto- Occipital Fasciculus
(R D)
−0.164 Paired Stimulus P2 Response 0.034
Right Cingulum—Cingulate Gyrus Portion
(R D)
−0.179 P300 Complex 0.02
Left Cingulum—Parahippocampal Portion
(R D)
−0.227 Oddball Ongoing High Frequency Activity −0.008
Right Superior Longitudinal Fasciculus (RD) −0.235 Latency −0.018
Forceps Minor (RD) −0.259 Antisaccade −0.295
FIGUR E | Scatterplot of component 1- weighted variables for the two sets (DTI RD, bio- factors). Scatterplot dots show each individual's subject
“performance” on each of the two component- weighted variables and are separated by diagnoses group: Persons with bipolar disorder (blue) and
healthy persons (gray). The x- axis is the component representing R D structure and y- axis is the component primarily driven by intrinsic EEG activ ity
and ERP strength related bio- factors (see Table4) with line fits per each subgroup included.
13995618, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/bdi.70010 by University Of Georgia Libraries, Wiley Online Library on [14/03/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
10 of 12 Bipolar Disorders, 2025
White matter microstructure was captured using DTI metrics,
FA and RD. Elevated RD results for the right CGC for persons
with BD complemented the reduced FA findings, emphasizing
the impacted connectivity of this tract. Higher RD and lower FA
may highlight the increased white matter disorganization for
this tract. The cingulum bundle is associated with emotion reg-
ulation, and impacted white matter microstructure may mani-
fest symptomatically with irregular emotional regulation in BD,
aligning with previous research [17].
Elevated RD values were noted in the bilateral anterior thalamic
radiation for persons with BD compared to healthy. This tract
connects portions of the prefrontal cortex and the thalamus and
supports multimodal processing. This finding aligns with liter-
ature noting BD- related abnormalities in this tract [34]. Higher
RD values were also noted bilaterally in the superior longitudi-
nal fasciculus (SLF) for persons with BD compared to healthy
persons. The SLF connects frontal cortices to occipital, parietal,
and temporal lobes and has been highlighted for memory, at-
tention, and emotion. Disrupted connectivity in this tract may
translate as deficits in visual processing, navigating environ-
ments, and motivational behaviors, as well as emotional and
cognitive dysfunction. Higher RD values were also noted in the
right inferior longitudinal fasciculus (ILF) for persons with BD
compared to healthy. The ILF connects occipital regions to an-
terior temporal cortical regions and is involved in visual cogni-
tion, as well as integrating visual and emotional information.
These results align with previous findings, with a meta- analysis
done by Vederine etal. [35] noting deviations in RD in the SLF,
ILF [36], and the forceps minor for persons with BD [6]. Elevated
RD values were also observed for bipolar persons compared to
healthy in the forceps major, a tract connecting lateral and me-
dial surfaces of the occipital lobe via the splenium of the corpus
callosum. Impacted connectivity may translate to compromised
visual and spatial processing.
The canonical correlation results between white matter struc-
ture and bio- factor data illustrate the usefulness of linking
across levels of analysis [37]. The weights of the RD component
(see Table4) indicate a relationship between less intact white
matter structure and especially higher intrinsic neural activity
and ERP strength (see Figure 7). B- SNIP described three psy-
chosis Biotypes (BT), with BT2 comprised of about 40% bipo-
lar psychosis cases [18]. The main features of BT2 in relation
to other B- SNIP Biotypes are modestly worse behavioral in-
hibition, modestly higher ERP amplitudes, and considerably
higher intrinsic, or background, brain activity. A combination
of modestly higher ERPs and considerably higher intrinsic ac-
tivity translates to poor signal- to- noise ratios, which can com-
promise signal processing and behavioral regulation [38]. The
relationship between those bio- factors and RD values possibly
yields information for developing a mechanistic understanding
of BD- relevant biomarker deviations.
There are not specific links to myelin integrity and intrinsic ac-
tivity in the literature, but there is indirect evidence of possi-
ble relevance. First, in a series of studies, Song and colleagues
[39] show that compromised myelin integrity may most clearly
be indicated by increased RD, and that changes in RD can
have behavioral consequences in human disease models [40].
Second, intact myelin may be important for modulating the
hyperexcitability of pyramidal neurons [41], and alternatively,
compromised myelin may yield pyramidal cell hyperexcitability.
Third, increased RD may be related to poor response inhibition
in a different psychopathology model with genetic overlap with
BD [42]. This brief survey provides a useful background for the
outcomes of the canonical correlation. The evidence is indirect
and requires further and specific exploration before judging
the veridicality of a hypothesis linking RD and specific physi-
ological deviations in BD or psychosis Biotypes. Nevertheless,
this information does illustrate a possible mechanism for a col-
lection of deviations observed in a specific subtype of serious
psychopathology.
Important considerations to make when reviewing the results
of this study include a constrained sample size, a limited trac-
tography atlas, and the varied physical health of subjects. The
inclusion of additional persons, especially in the bipolar with
and without psychosis subgroups, may help to determine more
critical features in white matter and intrinsic EEG activity and
ERP strength specific to psychosis in BD. Additionally, the
18- specified tractography atlas used in these analyses may not
be the most representative choice to capture all the differences
in white matter microstructure in BD or how they may relate
to EEG metrics. Including other tracts of interest may help
elucidate more about the relationship between white matter
and EEG. Factors like hypertension, hyperglycemia, hyperlip-
idemia, alcohol, tobacco, and drug use are all more prevalent
in bipolar patients and may potentially affect DTI metrics [43].
Future research should consider adjustments to DTI metrics
based on these variables.
This project illustrates the utility of objective and multimodal
approaches to understanding BD. White matter tract differences
were observed between BD and healthy groups. Additional re-
sults highlighted a key relationship between neural structure
and neural function such that reduced white matter microstruc-
ture may be associated with higher intrinsic neural activity and
higher ERP strength for persons with BD. Evaluating such re-
lationships is likely to provide more context to observed symp-
tomatology and shed more light on underlying neurobiological
processes of the disorder.
Acknowledgments
We would like to thank the participants from our studies, as well as the
research assistants and clinicians responsible for collecting and analyz-
ing clinical and biological data.
Ethics Statement
Research was conducted in compliance with U T Southwestern IRB at
each site, and informed consent was received from subjects all over the
age of 18, and data were anonymized and deidentified.
Conflicts of Interest
A.B., J.A.B., B.S. J., L.- Y.H., R .L.T., D.A.P., S.K.H., J.A.S. report no dis-
closures. E.I.: B- SNIP Diagnostics, Board of Managers; Consultant,
Janssen Pharmaceuticals; Consultant, Alkermes; Consultant, Karuna
Therapeutics. G.D.P.: B- SNIP Diagnostics, Board of Managers. C.A.T.:
B- SNIP Diagnostics, Board of Managers; Kynexis, Scientific Advisory
Board and receives a reta iner; Karuna Therapeutics, Scienti fic Advisory
13995618, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/bdi.70010 by University Of Georgia Libraries, Wiley Online Library on [14/03/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
11 of 12
Board and Own Stock. M.S.K.: B- SNIP Diagnostics, Board of Managers;
Advisor to Alkermes. S.K.K.: B- SNIP Diagnostics, Board of Managers.
E.S.G.: B- SNIP Diagnostics, Board of Managers; Consultant: Kynexis
Corporation. B.A.C.: B- SNIP Diagnostics, Board of Managers; Kynexis
Corporation, Scientific Advisor y Board. J.E.M.: B- SNIP Diagnostics,
Board of Managers.
Data Availability Statement
We elect not to share data.
References
1. American Psychiatric Association, ed., Diagnostic and Statistical
Manual of Mental Di sorders, 5th ed. (American Psychiatric Association,
2013).
2. B. A. Clementz, J. A. Sweeney, J. P. Hamm, etal., “Identification of
Distinct Psychosis Biotypes Using Brain- Based Biomarkers,” American
Journal of Psychiatr y 173, no. 4 (2016): 373–384.
3. K. S. Kendler, C. C. Masterson, and K. L. Davis, “Psychiatric Illness
in First- Degree Relatives of Patients With Paranoid Psychosis, Schizo-
phrenia and Medical Illness,” British Journal of Psychiatry 147, no. 5
(1985): 524–531.
4. E. Johnstone, C. D. Frith, T. J. Crow, J. Husband, and L. Kreel, “Cere-
bral Ventricular Si ze and Cognitive Impair ment in Chronic Schizophre -
nia,” Lancet 308 (1976): 924–926.
5. R. E. Roberts, E. J. Anderson, and M. Husain, “White Matter Micro-
structure and Cognitive Function,” Neuroscientist 19, no. 1 (2013): 8–15.
6. A. Manelis, A . Soehner, Y. O. Halchenko, etal., “W hite Matter Abnor-
malities in Adults With Bipolar Disorder Type- II and Unipolar Depres-
sion,” Scientific Reports 11 (2021): 7541.
7. D. A. Yurgelun- Todd, M. M. Silveri, S. A. Gruber, M. L. Rohan, and
P. J. Pimentel, “White Matter Abnormalities Observed in Bipolar Dis-
order: A Diffusion Tensor Imaging Study,” Bipolar Disorders 9, no. 5
(2007): 504–512.
8. P. Skudlarski, D. J. Schretlen, G. K. Thaker, etal., “Diffusion Tensor
Imaging White Matter Endophenotypes in Patients With Schizophrenia
or Psychotic Bipolar Disorder and Their Relatives,” American Journal of
Psychiatry 170, no. 8 (2013): 886–898.
9. D. F. Salisbury, M. E. Shenton, and R. W. McCarley, “P300 Topogra-
phy Differs in Schizophrenia and Manic Psychosis,” Biological Psychia-
try 45, no. 1 (1999): 98–106.
10. A. M. Morsel, M. Morrens, M. Dhar, and B. Sabbe, “Systematic Re-
view of Cognitive Event Related Potentials in Euthymic Bipolar Disor-
der,” Clinical Neurophysiology 129, no. 9 (2018): 1854–1865.
11. M. Wada, S. Kurose, T. Miyazaki, et al., “ The P300 Event- Related
Potential in Bipolar Disorder: A Sy stematic Review and Meta- A nalysis,”
Journal of Affective Disorders 256 (2019): 234–2 49.
12. D. A. Parker, R. L. Trotti, J. E. McDowell, et al., “Auditory Paired-
Stimuli Responses Across the Psychosis and Bipolar Spectrum and
Their Relationship to Clinical Features,” Biomarkers in Neuropsychia-
try 3 (2020): 100014, https:// doi. org/ 10. 1016/j. bionps. 2020. 100014.
13. J. P. Hamm, L. E. Ethridge, J. R. Shapiro, et al., “Family History of
Psychosis Moderates Early Auditor y Cortical Response Abnormalities in
Non- Psychotic Bipolar Disorder,” Bipolar Disorders 15 (2013): 774 –786.
14. S. L. Rossell, T. E. Rheenen, C. Groot, A. Gogos, A. O'Regan, and N.
R. Joshua, “Investigating Affective Prosody in Psychosis: A Study Using
the Comprehensive Affective Testing System,” Psychiatr y Research 210
(2013): 896–900.
15. K. Latalova, J. Prasko, T. Diveky, and H. Velartova, “Cognitive Im-
pairment in Bipolar Disorder,” Biomedical Papers of the Medical Faculty
of the University Palacky 155 (2011): 19–26.
16. C. A. Tamminga, E. I. Ivleva, M. S. Keshavan, etal., “Clinical Phe-
notypes of Psychosis in the Bipolar- Schizophrenia Network on Interme-
diate Phenotypes (B- SNIP),” American Journal of Philology 170, no. 11
(2013): 1263–1274.
17. J. A. Sweet, K. Gao, Z. Chen, etal., “Cingulum Bundle Connect ivity in
Treatment- Refractory Compared to Treatment- Responsive Patients With
Bipolar Disorder and Healthy Controls: A Tractography and Surgical Tar-
geting Analysis,” Journal of Neurosurgery 137, no. 3 (2022): 709–721.
18. D. A. Parker, R. Trotti, J. McDowell, etal., Differentiating Biomarker
Features and Familial Characteristics of B- SNIP Psychosis Biotypes (Re-
search Square, 202 4).
19. M. B. First, R. L. Spitzer, M. Gibbon, and J. B. W. Williams, Struc-
tured Clinical Interview for DSM- IV- TR Axis I Disorders, Research Ver-
sion (SCID- I/P) (CiNii Research, 2002).
20. M. B. First, R. L. Spitzer, M. Gibbon, and J. B. W. Willia ms, Structured
Clinical Intervie w for DSM- I V- TR Axis I Di sorders, Research Version. Non-
Patient. SCID- I/NP (New York State Psychiatric Institute, 2002).
21. J. A. Brown, B. S. Jackson, C. R. Burton, etal., “Reduced White Mat-
ter Microstructure in Bipolar Disorder With and Without Psychosis,”
Bipolar Disorders 23, no. 8 (2021): 801–809.
22. R. S. E. Keefe, P. D. Harvey, T. E. Goldberg, etal., “Norms and Stan-
dardization of the Brief Assessment of Cognition in Schizophrenia
(BACS),” Schizophrenia Research 102, no. 1 (2008): 108–115.
23. M. Y. Gotra, S. K. Hill, E. S. Gershon, et al., “Distinguishing Pat-
terns of Impairment on Inhibitory Control and General Cognitive Abil-
ity Among Bipolar With and Without Psychosis, Schizophrenia, and
Schizoaffective Disorder,” Schizophrenia Research 223 (2020): 148–157.
24. J. E. McDowell and B. A. Clementz, “Behavioral and Brain Imaging
Studies of Saccadic Performance in Schizophrenia,” Biological Psychol-
ogy 57, no. 1–3 (2001): 5–22.
25. O. Thomas, D. Parker, R. Trotti, et al., “Intrinsic Neural Activ-
ity Differences in Psychosis Biotypes: Findings From the Bipolar-
Schizophrenia Network on Intermediate Phenotypes (B- SNIP)
Consortium,” Biomarkers in Neuropsychiatry 1 (2019): 100002.
26. D. A. Parker, R. L. Trotti, J. E. McDowell, et al., “Auditory Oddball
Responses Across the Schizophrenia- Bipolar Spectrum and Their Re-
lationship to Cognitive and Clinical Features,” Amer ican Journal of
Psychiatry 178, no. 10 (2021): 952–964, https:// doi. org/ 10. 1176/ appi. ajp.
2021. 20071043.
27. J.- P. Fortin, D. Parker, B. Tunç, etal., “Harmonization of Multi- Site
Diffusion Tensor Imaging Data,” NeuroImage 161 (2017): 149–170 .
28. W. E. Johnson, C. Li, and A. Rabinovic, “Adjusting Batch Effects in
Microarray Expression Data Using Empirical Bayes Methods,” Biosta-
tistics 8, no. 1 (2007): 118–127.
29. D. J. Schaeffer, A. L. Rodrigue, C. R . Burton, etal., “White Matter
Structural Integrity Differs Between People With Schizophrenia and
Healthy Groups as a Function of Cognitive Control,” Schizophrenia Re-
search 169 (2015): 1–3.
30. J. Dukart, M. L . Schroeter, K. Mueller, and The Alzheimer's Disease
NI, “Age Correction in Dementia—Matching to a Healthy Brain,” PLoS
One 6, no. 7 (2011): 22193- 14.
31. B. A. Clementz, D. A. Parker, R. L. Trotti, etal., “Psychosis Biot ypes:
Replication and Validation From the B- SNIP Consortium,” Schizophre-
nia Bulletin 48, no. 1 (2022): 56 –68.
32. Z. V. Lambert, A. R. Wildt, and R. M. Durand, “Redundancy Analy-
sis: An Alternative to Canonical Correlation and Multivariate Multiple
Regression in Exploring Interset Associations,” Psychological Bulletin
104, no. 2 (1988): 282–289.
33. M. L. Knapp, J. M. Wiemann, and J. A. Daly, “Nonverbal Communi-
cation: Issues and Appraisal,” Human Communication Research 4, no.
3 (1978): 271–280.
13995618, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/bdi.70010 by University Of Georgia Libraries, Wiley Online Library on [14/03/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
12 of 12 Bipolar Disorders, 2025
34. R. Niida, B. Yamagata, A. Niida, A. Uechi, H. Matsuda, and M.
Mimura, “Aberrant Anterior Thalamic Radiation Structure in Bipolar
Disorder: A Diffusion Tensor Tractography Study,” Frontiers in Psychi-
atry 9 (2018): 522.
35. F. Vederine, M. Wessa, M. Leboyer, and J. Houenou, “A Meta-
Analysis of Whole- Brain Diffusion Tensor Imaging Studies in Bipolar
Disorder,” Progress in Neuro- Psychopharmacolog y & Biological Psychia-
try 35, no. 8 (2011): 1820–1826.
36. M. Barysheva, N. Jahanshad, L. Foland- Ross, L . L. Altshuler, and P.
M. Thompson, “White Matter Microstr uctural Abnormalities in Bipolar
Disorder: A Whole Brain Diffusion Tensor Imaging Study,” NeuroIm-
age: Clinical 2 (2013): 558–568.
37. A. L. Rodrigue, J. E. McDowell, N. Tandon, etal., “Multivariate Rela-
tionships Between Cognition and Brain A natomy Across the Psychosis
Spectrum,” Biological Psyc hiatry: Cognitive Neuroscience and Neuroimag-
ing 3, no. 12 (2018): 992–1002, https:// doi. org/ 10. 1016/j. bpsc. 2018. 03. 012.
38. M. E. Hudgens- Haney, L. E. Eth ridge, J. E. McDowell , etal., “Psycho-
sis Subgroups Differ in Intrinsic Neural Activity but Not Task- Specific
Processing,” Schizophrenia Research 195 (2018): 222–230, https:// doi.
org/ 10. 1016/j. schres. 2017. 08. 023.
39. R. T. Naismith, J. Xu, N. T. Tutlam, K. Trinkaus, A. H. Cross, and S.
Song, “Radial Diffusivity in Remote Optic Neuritis Discriminates Vi-
sual Outcomes,” Neurology 74, no. 21 (2010): 1702–1710.
40. J. Oh, S. Saidha, M. Chen, etal., “ Spinal Cord Quantitative MRI Dis-
criminates Between Disability Levels in Multiple Sclerosis,” Neurology
80, no. 6 (2013): 540 –547.
41. M. S. Hamada and M. H. P. Kole, “Myelin Loss and Axonal Ion
Channel Adaptations Associated With Gray Matter Neuronal Hyperex-
citabilit y,” Journal of Neuroscience 35, no. 18 (2015): 7272–7286.
42. D. Smullen, A. P. Bagshaw, L. Shalev, S. Tsafrir, T. Kolodny, and C.
Mevorach, “White Matter Properties in Fronto- Parietal Tracts Predict
Maladaptive Functional Activation and Deficient Response Inhibition
in ADHD,” bioRxiv 12, no. 4 (2023): 1–34.
43. P. Huang, Z. Shen, C . Wang, etal., “Altered White Mat ter Integrity in
Smokers Is Associated With Smok ing Cessation Outcomes,” Fr ontiers in
Human Neuroscience 11 (2017): 438.
13995618, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/bdi.70010 by University Of Georgia Libraries, Wiley Online Library on [14/03/2025]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
