Key PointsQuestion听
Are there objective, reproducible neural markers that can distinguish mania/hypomania from depression risk?
Findings听
In 3 independent samples comprising 299 young adults, neural response patterns differentially associated with mania/hypomania and depression risk were identified and replicated. Greater bilateral amygdala鈥搇eft amygdala functional connectivity was associated with greater mania/hypomania and depression risk, and greater bilateral ventrolateral prefrontal cortex鈥搑ight dorsolateral prefrontal cortex functional connectivity and greater right caudate deactivation were associated with greater mania/hypomania and depression risk, respectively.
Meaning听
Neural markers reliably associated with mania/hypomania and depression risk may help identify young adults at risk of bipolar disorder and provide treatment targets for early interventions.
Importance听
Mania/hypomania is the pathognomonic feature of bipolar disorder (BD). Established, reliable neural markers denoting mania/hypomania risk to help with early risk detection and diagnosis and guide the targeting of pathophysiologically informed interventions are lacking.
Objective听
To identify patterns of neural responses associated with lifetime mania/hypomania risk, the specificity of such neural responses to mania/hypomania risk vs depression risk, and the extent of replication of findings in 2 independent test samples.
Design, Setting, and Participants听
This cross-sectional study included 3 independent samples of young adults aged 18 to 30 years without BD or active substance use disorder within the past 3 months who were recruited from the community through advertising. Of 603 approached, 299 were ultimately included and underwent functional magnetic resonance imaging at the University of Pittsburgh, Pittsburgh, Pennsylvania, from July 2014 to May 2023.
Main Outcomes and Measures听
Activity and functional connectivity to approach-related emotions were examined using a region-of-interest mask supporting emotion processing and emotional regulation. The Mood Spectrum Self-Report assessed lifetime mania/hypomania risk and depression risk. In the discovery sample, elastic net regression models identified neural variables associated with mania/hypomania and depression risk; multivariable regression models identified the extent to which selected variables were significantly associated with each risk measure. Multivariable regression models then determined whether associations in the discovery sample replicated in both test samples.
Results听
A total of 299 participants were included. The discovery sample included 114 individuals (mean [SD] age, 21.60 [1.91] years; 80 female and 34 male); test sample 1, 103 individuals (mean [SD] age, 21.57 [2.09] years; 30 male and 73 female); and test sample 2, 82 individuals (mean [SD] age, 23.43 [2.86] years; 48 female, 29 male, and 5 nonbinary). Associations between neuroimaging variables and Mood Spectrum Self-Report measures were consistent across all 3 samples. Bilateral amygdala鈥搇eft amygdala functional connectivity and bilateral ventrolateral prefrontal cortex鈥搑ight dorsolateral prefrontal cortex functional connectivity were positively associated with mania/hypomania risk: discovery omnibus 蠂2鈥�=鈥�1671.7 (P鈥�<鈥�.001); test sample 1 omnibus 蠂2鈥�=鈥�1790.6 (P鈥�<鈥�.001); test sample 2 omnibus 蠂2鈥�=鈥�632.7 (P鈥�&濒迟;鈥�.001). Bilateral amygdala鈥搇eft amygdala functional connectivity and right caudate activity were positively associated and negatively associated with depression risk, respectively: discovery omnibus 蠂2鈥�=鈥�2566.2 (P鈥�<鈥�.001); test sample 1 omnibus 蠂2鈥�=鈥�2935.9 (P鈥�<鈥�.001); test sample 2 omnibus 蠂2鈥�=鈥�1004.5 (P鈥�&濒迟;鈥�.001).
Conclusions and Relevance听
In this study of young adults, greater interamygdala functional connectivity was associated with greater risk of both mania/hypomania and depression. By contrast, greater functional connectivity between ventral attention or salience and central executive networks and greater caudate deactivation were reliably associated with greater risk of mania/hypomania and depression, respectively. These replicated findings indicate promising neural markers distinguishing mania/hypomania鈥搒pecific risk from depression-specific risk and may provide neural targets to guide and monitor interventions for mania/hypomania and depression in at-risk individuals.
Bipolar disorder (BD), the pathognomonic feature of which is mania/hypomania,1 has peak onset in early adulthood.2 BD is often difficult to accurately diagnose, due to misreporting mania/hypomania, resulting in its frequent misclassification as unipolar depression.3 To facilitate earlier and more accurate BD diagnoses, there is a vital need for objective markers distinguishing risk of mania/hypomania from risk of depression.1 Identifying neural markers associated with risk for these symptoms can provide objective markers reflecting underlying pathophysiological processes and neural targets to guide and monitor interventions for BD.4
Three methodological approaches can be used to identify neural markers denoting risk of mania/hypomania. First, neuroimaging paradigms assessing neural responses underlying attention to approach-related emotional cues are especially useful for identifying markers associated with risk of mania/hypomania4 because of previous work highlighting behavioral approach system overactivity5,6 in mania/hypomania, resulting in attentional predisposition toward positive emotional stimuli7 (eg, happy faces and rewards8); emotional dysregulation9; and heightened positive affectivity, irritability, and anger.10
Second, studies should examine associations among neural responses in these contexts and comprehensive measures of mania/hypomania and depression risk. The Mood Spectrum Self-Report (MOODS-SR)11 measures lifetime vulnerability to mania/hypomania and depression, detecting subthreshold-level and threshold-level manifestations, with manic and depressive mood domain scales assessing the subtle behaviors over an individual鈥檚 lifetime that are associated with mania/hypomania and/or depression risk. By contrast, standard clinical rating scales assess current mania/hypomania or depression severity.11-13 The manic mood domain in particular discriminates between individuals with BD vs unipolar depression,11,13,14 highlighting the potential of this scale to screen for mania/hypomania and thus BD-specific risk in young adults on the BD spectrum, some of whom might already be diagnosed with unipolar depression.15,16
Additionally, studies should examine large-scale neural networks relevant to emotional dysregulation in approach-related contexts, including the central executive network (CEN), default mode network (DMN), salience network (SN), and ventral attention network (VAN).8,17,18 The CEN, centered on the dorsolateral prefrontal cortex (dlPFC) and caudate, supports attention-demanding tasks, decision making, and maintaining cognitive and emotional control.17,19 The DMN, centered on the medial prefrontal cortex (mPFC), precuneus, and posterior cingulate cortex, supports self-referential processing and introspection.20 The SN, centered on the (caudal) ventrolateral prefrontal cortex (vlPFC),21 dorsal anterior cingulate cortex (dACC), anterior insula, and amygdala, monitors, identifies, and integrates emotionally salient information19 with connections to reward (ventral striatum) and motor learning (putamen)22 regions. The VAN, centered on the (rostral) vlPFC and anterior insula, orients attention to emotionally salient stimuli and coactivates with CEN regions (eg, the dlPFC) when external stimuli refocus attention.23 Additionally, the orbitofrontal cortex (OFC) supports stimulus value encoding and updating,24 important for emotionally salient information processing.
Previous neuroimaging studies in individuals with BD reported amygdala and vlPFC hyperactivity to emotional stimuli,8,25 CEN hypoactivity26 and reduced CEN-amygdala functional connectivity during different cognitive and emotional-regulation paradigms26; SN hyperactivity to potential reward cues, notably in left caudal vlPFC8,27-29; and VAN or CEN18,30,31 and SN32 hyperconnectivity and altered DMN functional connectivity33,34 during resting state. Amygdala hyperactivity to positive emotional stimuli25; caudal left vlPFC hyperactivity to potential reward cues27; and elevated CEN, SN, and VAN functional connectivity in different contexts30-32 also distinguished individuals with BD from those with unipolar depression. Similarly, in a meta-analysis35 of functional neuroimaging studies in adult BD, we identified robust, reproducible, condition-dependent neural patterns characterizing BD, including altered left amygdala activity across emotional tasks, altered CEN activity across cognitive tasks, altered DMN resting-state functional connectivity, and altered CEN right caudate activity across all task types.
To our knowledge, no studies to date have combined all 3 of the above approaches. The few studies assessing neural markers of mania/hypomania risk in young adults reported similar findings of reward-related striatal and SN (insula) and emotional regulation鈥搑elated amygdala hyperactivity in individuals who were hypomania prone vs control individuals36,37; positive associations between mania/hypomania risk (measured by the MOODS-SR) and SN activity during reward processing38,39; and reduced resting-state functional connectivity between the ventral striatum and CEN regions in individuals at highest risk for BD.15 Studies in familial at-risk populations (eg, offspring of individuals affected by BD) and youth affected by BD similarly reported altered amygdala, striatal, CEN, and VAN or SN activity and altered amygdala-VAN and amygdala-SN functional connectivity during emotion processing and emotional regulation.40-43 Yet, although childhood-onset BD is a more severe form of BD,44 it is less common than adult-onset BD.45 Thus, there remains a critical need for studies to examine young adults at risk of mania/hypomania.35 There is also a need to address the larger replication crisis in brain-behavior research46 and replicate neuroimaging findings in independent samples47,48 to yield reliable, robust neural markers of BD risk.38
The goal of the present study was to identify neural markers of mania/hypomania vs depression risk in young adults. We first aimed to identify neural markers of mania/hypomania risk during approach-related emotion processing using the MOODS-SR manic mood measure of mania/hypomania risk in a discovery sample of young adults aged 18 to 30 years recruited from the general community across a range of such risk. Based on the previous findings described above, we hypothesized that greater mania/hypomania risk would be associated with amygdala and SN hyperactivity, CEN hypoactivity, CEN-amygdala hypoconnectivity, altered DMN functional connectivity, and VAN-CEN hyperconnectivity to approach-related emotional cues (hypothesis 1). We next aimed to determine whether associations found were specific to mania/hypomania risk. We hypothesized that findings would be mania/hypomania specific, and not common to depression risk (hypothesis 2). We lastly aimed to determine whether risk-specific associations could be independently replicated. We hypothesized that these associations would replicate in 2 independent young adult test samples (hypothesis 3).
Participants and Measures
A total of 299 young adult participants composed 3 independent samples recruited across a range of subsyndromal-syndromal affective and anxiety psychopathology (excluding BD and active substance use disorder within the past 3 months). Participants were recruited through student counseling centers, participant registries, and community advertisements. The University of Pittsburgh institutional review board approved this study; all participants gave written informed consent. For additional exclusion criteria, medication, and power calculations, see the eMethods in Supplement 1.
The discovery sample included 114 individuals (mean [SD] age, 21.60 [1.91] years; 80 female and 34 male; 36 with lifetime diagnoses of major depressive disorder, attention-deficit/hyperactivity disorder, or anxiety disorders and 78 without). Test sample 1 included 103 individuals (mean [SD] age, 21.57 [2.09] years; 30 male and 73 female; 47 with psychiatric diagnoses and 56 without). Test sample 2 included 82 individuals (mean [SD] age, 23.43 [2.86] years; 48 female, 29 male, and 5 nonbinary; 19 with psychiatric diagnoses and 63 without) (Table 1).
Participants鈥� lifetime mania/hypomania and depression risk were assessed by the MOODS-SR11 mood domains. The manic mood domain assessed euphoria, inflated self-esteem, mixed instability or irritability, creativity, sociability or extraversion, and wastefulness or recklessness. The depressive mood domain assessed depressive mood and substance use鈥搑elated depression.13 While other clinical scales assessed present symptom severity as part of a larger study (eMethods in Supplement 1), they were not the focus of this study.
Functional Magnetic Resonance Imaging Task
Participants completed a facial emotion processing task (eFigure 1 in Supplement 1). The main stimulus contrast was approach-related emotional expressions (angry and happy) vs implicit baseline. For functional magnetic resonance imaging task, acquisition parameters, and preprocessing details, see the eMethods in Supplement 1.
Activity and Functional Connectivity
Given extant findings highlighting functional alterations in BD predominantly in the DMN, CEN, VAN, and SN prefrontal cortical regions, an anatomically defined region-of-interest mask included the following bilateral regions from these a priori networks: DMN mPFC (Brodmann area [BA] 10), CEN dlPFC (BA9 and BA46), VAN and SN vlPFC (BA47), SN vACC (BA24), and dACC (BA32), as well as OFC (BA11), and other regions connected with the SN and CEN supporting emotional salience, cognition, and motor learning鈥攊e, bilateral amygdala, insula, and striatum (caudate, putamen, and ventral striatum) (spheres at Montreal Neurologic Institute [MNI] coordinates 9, 9, 鈭�8 and 鈭�9, 9, 鈭�8; radius鈥�=鈥�8 mm). Parameter estimates of significant clusters of activity within this mask (familywise error P鈥�&濒迟;鈥�.05; cluster size, k鈥�>鈥�20 voxels) to our main stimulus contrast were extracted in SPM12 (Statistical Parametric Mapping; The Wellcome Trust Centre for Neuroimaging).
Generalized psychophysiological interaction50 calculated significant functional connectivity between anatomically defined bilateral seed regions representing the CEN (dlPFC and caudate), SN (dACC, amygdala, and putamen), and VAN and SN (vlPFC) and targets within the rest of the above region-of-interest mask for the main contrast (familywise error P鈥�&濒迟;鈥�.05; k鈥�>鈥�20 voxels). In each sample, we separately identified and extracted parameter estimates of approach-related activity and functional connectivity within the above region-of-interest mask.
Associations With Mania/Hypomania and Depression Risk
To test hypotheses 1 and 2 in the discovery sample, elastic net鈥損enalized least squares regression for variable selection with 10-fold cross-validation was performed using GLMNET in R version 4.0 (R Foundation) and the appropriate regression family to identify nonzero independent variables in 2 separate risk models: 1 for MOODS-SR mania/hypomania and 1 for MOODS-SR depression (dependent variables). Significant neural measures (activity and functional connectivity) and demographic variables (age and gender) were independent variables. Model coefficients were selected at 位 minimum.
Next, in the discovery sample, using SPSS version 27 (IBM) and the 2 separate risk models, the appropriate regression family model assessed the magnitude of the associations between nonzero independent variables and each MOODS-SR DV. Results were corrected for multiple comparisons (false discovery rate [FDR]鈥揳djusted P鈥�<鈥�.05) across both models within each sample.
To further determine each model鈥檚 specificity to mania/hypomania or depression risk, all nonzero coefficients from both models were included as independent variables in additional mania/hypomania and depression risk models.
To test hypothesis 3, in 2 independent test samples, SPSS regression models using the nonzero independent variables identified from the discovery sample determined whether independent variable鈥揹ependent variable associations replicated in the 2 test samples. Standard cross-validation then evaluated the models鈥� predictive performance in each sample51 (eMethods in Supplement 1). Additional risk models were then performed in each test sample as above.
Post hoc sensitivity tests of the above SPSS regression model procedures for the 3 hypotheses were performed in subsets of participants from each sample who were unmedicated and without lifetime diagnoses of major depressive disorder or attention-deficit/hyperactivity disorder, as these disorders can be misdiagnosed as BD3,52 and can precede BD and other psychiatric disorders.52-54 Exploratory analyses assessed neural responses to all facial emotions (angry, happy, sad, and fearful) vs implicit baseline to test whether our main findings were specific to approach-related emotions.
The patterns of approach-related neural activity and functional connectivity in the 2 test samples were mostly consistent with that shown in the discovery sample (eTable 2 in Supplement 1), which allowed the 2 risk models generated in the discovery sample to be tested for replication in the test samples. Given that the MOODS-SR is a nonnegative count measure and its distributions in each sample were nonnormal (ie, zero inflated and positively skewed) (eTable 1 in Supplement 1), we used Poisson regression models for all analyses.55-57
Replicated Associations With Mania/Hypomania Risk
In the discovery sample, we first identified neural patterns associated with mania/hypomania risk. Elastic net variable selection after cross-validation using the Poisson family revealed that age, left dlPFC activity, right vlPFC activity, amygdala鈥搇eft amygdala functional connectivity, vlPFC鈥搑ight dlPFC functional connectivity, and dACC鈥搇eft mPFC functional connectivity to approach-related emotions were associated with mania/hypomania risk (Table 2). A Poisson log-link regression model then revealed that left dlPFC (FDR Q鈥�=鈥�0.003) and right vlPFC (FDR Q鈥�=鈥�0.007) activity were negatively associated with mania/hypomania risk, and amygdala鈥搇eft amygdala functional connectivity (Figures 1A and 2A), vlPFC鈥搑ight dlPFC functional connectivity (Figures 1B and 2A), and dACC鈥搇eft mPFC functional connectivity (FDR Q鈥�<鈥�0.001) were positively associated with mania/hypomania risk (Table 2). Age was not significantly associated with mania/hypomania risk.
Two findings replicated in both test samples. Amygdala鈥搇eft amygdala functional connectivity (Figure 1A) and vlPFC鈥搑ight dlPFC functional connectivity (Figure 1B) were positively associated with mania/hypomania risk.
In test sample 1, a Poisson log-link regression model revealed that amygdala鈥搇eft amygdala functional connectivity (Figures 1A and 2B) and vlPFC鈥搑ight dlPFC functional connectivity (FDR Q鈥�<鈥�0.001) (Figures 1B and 2B) were positively associated with mania/hypomania risk. Age and dACC鈥搇eft mPFC functional connectivity were not significantly associated with mania/hypomania risk (Table 2).
In test sample 2, a Poisson log-link regression model revealed that amygdala鈥搇eft amygdala functional connectivity (FDR Q鈥�<鈥�0.001) (Figures 1A and 2C), vlPFC鈥搑ight dlPFC functional connectivity (FDR Q鈥�=鈥�0.006) (Figures 1B and 2C), and dACC鈥搇eft mPFC functional connectivity (FDR Q鈥�=鈥�0.009) were positively associated with mania/hypomania risk. Age was not significantly associated with mania/hypomania risk (Table 2).
Left dlPFC and right vlPFC activity were not observed in either test sample and thus were not included in the above models. Standard cross-validation for mania/hypomania indicated similar predictive performance: discovery sample鈥�=鈥�0.75, test sample 1鈥�=鈥�0.74, and test sample 2鈥�=鈥�0.84. For additional mania/hypomania risk models, see the eResults and eTable 3 in Supplement 1.
Replicated Associations With Depression Risk
In the discovery sample, we tested whether the above neural patterns were common to depression risk. Elastic net variable selection after cross-validation using the Poisson family revealed that age, right vlPFC activity, right caudate activity, amygdala鈥搇eft amygdala functional connectivity, putamen鈥搑ight mPFC functional connectivity, putamen鈥搇eft dACC functional connectivity, dlPFC鈥搑ight dlPFC functional connectivity, and dACC鈥搇eft dlPFC functional connectivity to approach-related emotions were associated with depression risk (Table 3). A Poisson log-link regression model then revealed that age (FDR Q鈥�=鈥�0.01), right vlPFC activity, right caudate activity (Figures 1C and 2A), and dACC鈥搇eft dlPFC functional connectivity (FDR Q鈥�<鈥�0.001) were negatively associated with depression risk, and amygdala鈥搇eft amygdala functional connectivity (Figures 1A and 2A) and dlPFC鈥搑ight dlPFC functional connectivity (FDR Q鈥�<鈥�0.001) were positively associated with depression risk (Table 3). Putamen鈥搑ight mPFC functional connectivity and putamen鈥搇eft dACC functional connectivity were not significantly associated with depression risk.
Two findings replicated in both test samples. Amygdala鈥搇eft amygdala functional connectivity (Figure 1A) and right caudate activity (Figure 1C) were positively and negatively associated with depression risk, respectively.
In test sample 1, a Poisson log-link regression model revealed that right caudate activity (Figures 1C and 2B) was negatively associated with depression risk, and amygdala鈥搇eft amygdala functional connectivity (Figures 1A and 2B), dlPFC鈥搑ight dlPFC functional connectivity, and putamen鈥搇eft dACC functional connectivity (FDR Q鈥�<鈥�0.001) were positively associated with depression risk (Table 3). Age and dACC鈥搇eft dlPFC functional connectivity were not significantly associated with depression risk. Right vlPFC activity and putamen鈥搑ight mPFC functional connectivity were not observed in test sample 1 and thus were not included in the model.
In test sample 2, a Poisson log-link regression model revealed that right caudate activity (Figures 1C and 2C) and putamen鈥搇eft dACC functional connectivity were negatively associated with depression risk, and age, amygdala鈥搇eft amygdala functional connectivity (Figures 1A and 2C), dACC鈥搇eft dlPFC functional connectivity, and putamen鈥搑ight mPFC functional connectivity (FDR Q鈥�<鈥�0.001) were positively associated with depression risk. dlPFC鈥搑ight dlPFC functional connectivity was not significantly associated with depression risk (Table 3). Right vlPFC activity was not observed in test sample 2 and thus was not included in the model. Cross-validation for depression revealed similar predictive performance: discovery sample鈥�=鈥�0.85, test sample 1鈥�=鈥�1.14, and test sample 2鈥�=鈥�1.08. For additional depression risk models, see eResults and eTable 4 in Supplement 1.
The replicated associations were largely consistent when excluding medicated individuals and individuals with major depressive disorder and attention-deficit/hyperactivity disorder. However, they were not fully replicated in all facial-emotion exploratory analyses, highlighting the specificity of risk markers to approach-related emotions (eResults and eTables 5-13 in Supplement 1).
In this cross-sectional study, we aimed to identify reliable neural markers distinguishing mania/hypomania from depression risk in young adulthood, when psychiatric disorders such as BD often manifest,2 via a novel combination of 3 approaches: using an approach-related emotion-processing task, examining large-scale neural networks, and the MOODS-SR to measure mania/hypomania and depression vulnerability. Our findings support our first hypothesis, especially associations between greater amygdala (SN) activity and VAN-CEN functional connectivity and greater mania/hypomania risk; our second hypothesis that distinct neural response patterns would be associated with mania/hypomania vs depression risk; and our third hypothesis that these findings would replicate in 2 independent samples. Specifically, we showed, in 3 independent young adult samples with varying degrees of psychopathology, replicated associations between elevated bilateral amygdala鈥搇eft amygdala functional connectivity and mania/hypomania and depression risk; greater VAN-CEN (bilateral vlPFC鈥搑ight dlPFC) functional connectivity and greater mania/hypomania risk; and greater CEN (right caudate) deactivation and greater depression risk.
The amygdala finding parallels animal58-60 and human imaging61-63 studies showing synchronous interamygdala activation and functional connectivity and interamygdala coupling during emotion processing in nonclinical populations.64,65 In the present study, the left amygdala target extended beyond the amygdala seed mask, possibly extending to the DMN parahippocampal gyrus in the test samples (eFigure 2 in Supplement 1), but the key associations between interamygdala functional connectivity and mania/hypomania and depression risk parallel the above findings. These findings highlight interamygdala functional coupling, possibly reflecting greater attribution of salience to approach-related emotional stimuli, as a potential marker of broader mood disorder risk. Interestingly, bilateral amygdala鈥搇eft amygdala functional connectivity to all facial emotions was associated with depression risk, but not mania/hypomania risk, in the discovery sample (missing significance in both test samples). Thus, depression risk might be associated with greater attribution of salience to all emotions, with mania/hypomania risk more specifically associated with heightened salience to approach-related emotions, paralleling findings of amygdala hyperactivity to positive emotional stimuli in BD but not unipolar depresson.25
By contrast, greater VAN-CEN engagement, indicating greater attention or sensitivity to this specific approach-related emotional context, uniquely characterized mania/hypomania risk, paralleling findings highlighting associations among attentional predispositions toward negative66 and positive emotional stimuli,7,10 elevated behavioral approach system,5,6 and mania/hypomania risk. Our reproducible positive associations between depression risk and caudate deactivation are consistent with reduced CEN function in various contexts shown previously in depression,32,67,68 which in turn can be associated with lower emotional regulation capacity.69,70 Future studies should further examine these associations and other VAN or CEN regions when testing neural models of BD, as functional alterations in these networks can not only inform studies aiming to replicate our findings in other populations at risk of BD but also help identify prodromal markers in individuals with established BD.
Not all activity or functional connectivity in the discovery sample replicated in both test samples. Bilateral dACC鈥搇eft mPFC (SN-DMN) functional connectivity was positively associated with mania/hypomania risk in the discovery sample and test sample 2 only, paralleling previous studies in BD showing elevated DMN functional connectivity with other neural networks and thought to result in compromised functioning of these other networks.71 Bilateral dlPFC鈥搑ight dlPFC (CEN-CEN) functional connectivity was positively associated with depression risk in the discovery sample and test sample 1 only, potentially reflecting inefficient, compensatory CEN recruitment and paralleling our previous findings associating elevated CEN activity with depression.67 Bilateral dACC鈥搇eft dlPFC (SN-CEN) functional connectivity and age associations with depression risk in opposite directions in the discovery and test sample 2 and bilateral putamen鈥揹ACC (SN-SN) associations with depression risk in opposite directions in the test samples further suggest associations between SN and CEN or motor network functional connectivity, age, and depression risk, although these inconsistent findings should be interpreted with caution, given the directionality differences.
In additional risk models including all nonzero coefficients associated with mania/hypomania and depression risk, caudate deactivation was associated with mania/hypomania and depression risk in all 3 samples whereas vlPFC-dlPFC functional connectivity was associated with depression risk in test sample 1 only, (and these associations were negative, rather than positive, in the other 2 samples), and remained a significant marker of mania/hypomania risk in 2 samples. Caudate deactivation was thus a less specific marker of depression risk than vlPFC-dlPFC functional connectivity was of mania/hypomania risk.
This study had limitations. While a strength was the replication of findings in independent adequately powered samples (eMethods in Supplement 1), future replications are needed. Our findings did not support all (especially CEN-related) network-based hypotheses associated with mania/hypomania risk. Future studies can use tasks that engage the CEN more than our present paradigm. While not all neural measures were identified in all 3 samples, potentially reflecting acquisition parameter differences, we identified robustly replicated neural markers of mania/hypomania and depression risk. While our samples differed on demographic variables and MOODS-SR scores (Table 1), the findings replicated across samples and when using different data analytic pipelines. Sensitivity analyses confirmed that the significant associations generalized to unmedicated individuals and those without major depressive disorder or attention-deficit/hyperactivity disorder. We did not examine neural markers of state anxiety and current mania/hypomania and depression severity, as these were correlated with MOODS-SR measures (eTable 14 and eFigures 3-5 in Supplement 1). We used the MOODS-SR to comprehensively measure future mania/hypomania risk. Future studies can aim to identify neural markers associated with future manic/hypomanic episodes.
The findings in this study show, in 3 independent samples, robust associations between specific neural markers and mania/hypomania and depression risk. These replicated findings indicate promising neural markers for distinguishing mania/hypomania from depression risk and provide neural targets to guide and monitor interventions in individuals at risk for BD and other affective disorders.
Accepted for Publication: August 24, 2023.
Published Online: November 1, 2023. doi:10.1001/jamapsychiatry.2023.4150
Open Access: This is an open access article distributed under the terms of the CC-BY License. 漏 2023 Schumer MC et al. JAMA Psychiatry.
Corresponding Author: Maya C. Schumer, BS, Department of Psychiatry, University of Pittsburgh School of Medicine, 121 Meyran Ave, 203 Loeffler Bldg, Pittsburgh, PA 15213 (schumerm@upmc.edu).
Author Contributions: Ms Schumer and Dr Phillips had full access to all the data in the study and take full responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Schumer, Bertocci, Phillips.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Schumer, Bertocci, Skeba, Phillips.
Critical review of the manuscript for important intellectual content: All authors.
Statistical analysis: Schumer, Bertocci, Chase, Phillips.
Obtained funding: Phillips.
Administrative, technical, or material support: Aslam, Graur, Bebko, Stiffler, Skeba, Brady, Benjamin, Wang, Chase, Phillips.
Supervision: Bertocci, Graur, Phillips.
Conflict of Interest Disclosures: Ms Schumer reported other from the Association for Women in Science (doctoral dissertation scholarship) during the conduct of the study. Dr Bertocci reported grants from Brain and Behavior Research Foundation during the conduct of the study. Dr Phillips reported grants from the National Institute of Mental Health and other from the Pittsburgh Foundation (endowment) during the conduct of the study. No other disclosures were reported.
Funding/Support: This study was supported by the National Institute of Mental Health (R01MH100041 and R37MH100041), the Pittsburgh Foundation, the Brain and Behavior Research Foundation, and the Association for Women in Science. This research was also supported in part by the University of Pittsburgh Center for Research Computing (RRID:SCR022735) through resources provided. Specifically, this work used the High Throughput Computing cluster, which is supported by National Institutes of Health award number S10OD028483.
Role of the Funder/Sponsor: The Pittsburgh Foundation, Brain and Behavior Research Foundation, and Association for Women in Science partially funded salary support for investigators responsible for the design and conduct of the study, collection, management, and analysis. National Institute of Mental Health grants partially or fully supported salaries for investigators, students, and staff responsible for the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Data Sharing Statement: See Supplement 2.
1.American Psychiatric Association. 听Diagnostic and Statistical Manual of Mental Disorders (DSM-5). American Psychiatric Association; 2013.
2.Kessler
听RC锘�, Berglund
听P锘�, Demler
听O锘�, Jin
听R锘�, Merikangas
听KR锘�, Walters
听EE锘�. 听Lifetime prevalence and age-of-onset distributions of DSM-IV disorders in the National Comorbidity Survey Replication.听锘� 听Arch Gen Psychiatry. 2005;62(6):593-602. doi:
3.Bowden
听CL锘�. 听Strategies to reduce misdiagnosis of bipolar depression.听锘� 听Psychiatr Serv. 2001;52(1):51-55. doi:
4.Cardoso de Almeida
听JR锘�, Phillips
听ML锘�. 听Distinguishing between unipolar depression and bipolar depression: current and future clinical and neuroimaging perspectives.听锘� 听Biol Psychiatry. 2013;73(2):111-118. doi:
5.Johnson
听SL锘�, Edge
听MD锘�, Holmes
听MK锘�, Carver
听CS锘�. 听The behavioral activation system and mania.听锘� 听Annu Rev Clin Psychol. 2012;8:243-267. doi:
6.Urosevi膰
听S锘�, Abramson
听LY锘�, Harmon-Jones
听E锘�, Alloy
听LB锘�. 听Dysregulation of the behavioral approach system (BAS) in bipolar spectrum disorders: review of theory and evidence.听锘� 听Clin Psychol Rev. 2008;28(7):1188-1205. doi:
7.Gruber
听J锘�. 听Can feeling too good be bad?: positive emotion persistence (PEP) in bipolar disorder.听锘� 听Curr Dir Psychol Sci. 2011;20(4):217-221. doi:
8.Phillips
听ML锘�, Swartz
听HA锘�. 听A critical appraisal of neuroimaging studies of bipolar disorder: toward a new conceptualization of underlying neural circuitry and a road map for future research.听锘� 听Am J Psychiatry. 2014;171(8):829-843. doi:
9.Bigot
听M锘�, Alonso
听M锘�, Houenou
听J锘�,
听et al. 听An emotional-response model of bipolar disorders integrating recent findings on amygdala circuits.听锘� 听Neurosci Biobehav Rev. 2020;118:358-366. doi:
10.Gruber
听J锘�, Johnson
听SL锘�, Oveis
听C锘�, Keltner
听D锘�. 听Risk for mania and positive emotional responding: too much of a good thing?听锘� 听贰尘辞迟颈辞苍. 2008;8(1):23-33. doi:
11.Dell鈥橭sso
听L锘�, Armani
听A锘�, Rucci
听P锘�,
听et al. 听Measuring mood spectrum: comparison of interview (SCI-MOODS) and self-report (MOODS-SR) instruments.听锘� 听Compr Psychiatry. 2002;43(1):69-73. doi:
12.Mineo
听L锘�, Rodolico
听A锘�, Spedicato
听GA锘�,
听et al. 听Exploration of mood spectrum symptoms during a major depressive episode: the impact of contrapolarity鈥攔esults from a transdiagnostic cluster analysis on an Italian sample of unipolar and bipolar patients.听锘� 听Eur Psychiatry. 2022;65(1):e30. doi:
13.Fagiolini
听A锘�, Dell鈥檕sso
听L锘�, Pini
听S锘�,
听et al. 听Validity and reliability of a new instrument for assessing mood symptomatology: the Structured Clinical Interview for Mood Spectrum (SCI-MOODS).听锘� 听Int J Methods Psychiatr Res. 1999;8(2):71-82. doi:
14.Ioannou
听M锘�, Dellepiane
听M锘�, Benvenuti
听A锘�, Feloukatzis
听K锘�, Skondra
听N锘�, Steingrimsson
听S锘�. 听Psychometric evaluation of a 33-item subset of MOODS-SR for distinguishing bipolar disorder.听锘� 听Eur Psychiatry. 2016;33(S1):s223-s223. doi:
15.Kling
听LR锘�, Bessette
听KL锘�, DelDonno
听SR锘�,
听et al. 听Cluster analysis with MOODS-SR illustrates a potential bipolar disorder risk phenotype in young adults with remitted major depressive disorder.听锘� 听Bipolar Disord. 2018;20(8):697-707. doi:
16.Cassano
听GB锘�, Rucci
听P锘�, Benvenuti
听A锘�,
听et al. 听The role of psychomotor activation in discriminating unipolar from bipolar disorders: a classification-tree analysis.听锘� 听J Clin Psychiatry. 2012;73(1):22-28. doi:
17.Bi
听B锘�, Che
听D锘�, Bai
听Y锘�. 听Neural network of bipolar disorder: toward integration of neuroimaging and neurocircuit-based treatment strategies.听锘� 听Transl Psychiatry. 2022;12(1):143. doi:
18.Gong
听J锘�, Wang
听J锘�, Chen
听P锘�,
听et al. 听Large-scale network abnormality in bipolar disorder: a multimodal meta-analysis of resting-state functional and structural magnetic resonance imaging studies.听锘� 听J Affect Disord. 2021;292:9-20. doi:
19.Menon
听V锘�. 听Large-scale brain networks and psychopathology: a unifying triple network model.听锘� 听Trends Cogn Sci. 2011;15(10):483-506. doi:
20.Raichle
听ME锘�. 听The brain鈥檚 default mode network.听锘� 听Annu Rev Neurosci. 2015;38(1):433-447. doi:
21.Trambaiolli
听LR锘�, Peng
听X锘�, Lehman
听JF锘�,
听et al. 听Anatomical and functional connectivity support the existence of a salience network node within the caudal ventrolateral prefrontal cortex.听锘� 听别尝颈蹿别. 2022;11:e76334. doi:
22.Peters
听SK锘�, Dunlop
听K锘�, Downar
听J锘�. 听Cortico-striatal-thalamic loop circuits of the salience network: a central pathway in psychiatric disease and treatment.听锘� 听Front Syst Neurosci. 2016;10:104. doi:
23.Viviani
听R锘�. 听贰尘辞迟颈辞苍 regulation, attention to emotion, and the ventral attentional network.听锘� 听Front Hum Neurosci. 2013;7:746. doi:
24.Robles Aguirre
听FA锘�, Hern谩ndez Gonz谩lez
听M锘�, P茅rez Hern谩ndez
听M锘�. del Carmen Rodr铆guez Flores
听T锘�, Guevara
听M脕锘�. 听Inquiry of the orbitofrontal cortex role in incentive learning: an artificial neural networks simulation.听锘� 听Cogn Syst Res. 2023;78:14-22. doi:
25.Grotegerd
听D锘�, Stuhrmann
听A锘�, Kugel
听H锘�,
听et al. 听Amygdala excitability to subliminally presented emotional faces distinguishes unipolar and bipolar depression: an fMRI and pattern classification study.听锘� 听Hum Brain Mapp. 2014;35(7):2995-3007. doi:
26.Chase
听HW锘�, Phillips
听ML锘�. 听Elucidating neural network functional connectivity abnormalities in bipolar disorder: toward a harmonized methodological approach.听锘� 听Biol Psychiatry Cogn Neurosci Neuroimaging. 2016;1(3):288-298. doi:
27.Chase
听HW锘�, Nusslock
听R锘�, Almeida
听JR锘�, Forbes
听EE锘�, LaBarbara
听EJ锘�, Phillips
听ML锘�. 听Dissociable patterns of abnormal frontal cortical activation during anticipation of an uncertain reward or loss in bipolar versus major depression.听锘� 听Bipolar Disord. 2013;15(8):839-854. doi:
28.Nusslock
听R锘�, Almeida
听JR锘�, Forbes
听EE锘�,
听et al. 听Waiting to win: elevated striatal and orbitofrontal cortical activity during reward anticipation in euthymic bipolar disorder adults.听锘� 听Bipolar Disord. 2012;14(3):249-260. doi:
29.Caseras
听X锘�, Lawrence
听NS锘�, Murphy
听K锘�, Wise
听RG锘�, Phillips
听ML锘�. 听Ventral striatum activity in response to reward: differences between bipolar I and II disorders.听锘� 听Am J Psychiatry. 2013;170(5):533-541. doi:
30.Goya-Maldonado
听R锘�, Brodmann
听K锘�, Keil
听M锘�, Trost
听S锘�, Dechent
听P锘�, Gruber
听O锘�. 听Differentiating unipolar and bipolar depression by alterations in large-scale brain networks.听锘� 听Hum Brain Mapp. 2016;37(2):808-818. doi:
31.He
听H锘�, Yu
听Q锘�, Du
听Y锘�,
听et al. 听Resting-state functional network connectivity in prefrontal regions differs between unmedicated patients with bipolar and major depressive disorders.听锘� 听J Affect Disord. 2016;190:483-493. doi:
32.Siegel-Ramsay
听JE锘�, Bertocci
听MA锘�, Wu
听B锘�, Phillips
听ML锘�, Strakowski
听SM锘�, Almeida
听JRC锘�. 听Distinguishing between depression in bipolar disorder and unipolar depression using magnetic resonance imaging: a systematic review.听锘� 听Bipolar Disord. 2022;24(5):474-498. doi:
33.Zovetti
听N锘�, Rossetti
听MG锘�, Perlini
听C锘�,
听et al. 听Default mode network activity in bipolar disorder.听锘� 听Epidemiol Psychiatr Sci. 2020;29:e166. doi:
34.Xue
听C锘�, Zhang
听X锘�, Cao
听P锘�,
听et al. 听Evidence of functional abnormalities in the default mode network in bipolar depression: A coordinate-based activation likelihood estimation meta-analysis.听锘� 听J Affect Disord. 2023;326:96-104. doi:
35.Schumer
听MC锘�, Chase
听HW锘�, Rozovsky
听R锘�, Eickhoff
听SB锘�, Phillips
听ML锘�. 听Prefrontal, parietal, and limbic condition-dependent differences in bipolar disorder: a large-scale meta-analysis of functional neuroimaging studies.听锘� 听Mol Psychiatry. Published online February 13, 2023. doi:
36.Heissler
听J锘�, Kanske
听P锘�, Sch枚nfelder
听S锘�, Wessa
听M锘�. 听Inefficiency of emotion regulation as vulnerability marker for bipolar disorder: evidence from healthy individuals with hypomanic personality.听锘� 听J Affect Disord. 2014;152-154:83-90. doi:
37.O鈥橲ullivan
听N锘�, Szczepanowski
听R锘�, El-Deredy
听W锘�, Mason
听L锘�, Bentall
听RP锘�. 听fMRI evidence of a relationship between hypomania and both increased goal-sensitivity and positive outcome-expectancy bias.听锘� 听狈别耻谤辞辫蝉测肠丑辞濒辞驳颈补. 2011;49(10):2825-2835. doi:
38.Edmiston
听EK锘�, Fournier
听JC锘�, Chase
听HW锘�,
听et al. 听Assessing relationships among impulsive sensation seeking, reward circuitry activity, and risk for psychopathology: a functional magnetic resonance imaging replication and extension study.听锘� 听Biol Psychiatry Cogn Neurosci Neuroimaging. 2020;5(7):660-668. doi:
39.de Oliveira
听L锘�, Portugal
听LCL锘�, Pereira
听M锘�,
听et al. 听Predicting bipolar disorder risk factors in distressed young adults from patterns of brain activation to reward: a machine learning approach.听锘� 听Biol Psychiatry Cogn Neurosci Neuroimaging. 2019;4(8):726-733. doi:
40.Ladouceur
听CD锘�, Diwadkar
听VA锘�, White
听R锘�,
听et al. 听Fronto-limbic function in unaffected offspring at familial risk for bipolar disorder during an emotional working memory paradigm.听锘� 听Dev Cogn Neurosci. 2013;5:185-196. doi:
41.Ladouceur
听CD锘�, Farchione
听T锘�, Diwadkar
听V锘�,
听et al. 听Differential patterns of abnormal activity and connectivity in the amygdala-prefrontal circuitry in bipolar-I and bipolar-NOS youth.听锘� 听J Am Acad Child Adolesc Psychiatry. 2011;50(12):1275-89.e2. doi:
42.Manelis
听A锘�, Ladouceur
听CD锘�, Graur
听S锘�,
听et al. 听Altered amygdala-prefrontal response to facial emotion in offspring of parents with bipolar disorder.听锘� 听叠谤补颈苍. 2015;138(Pt 9):2777-2790. doi:
43.Nery
听FG锘�, Welge
听JA锘�, Fleck
听D锘�,
听et al. 听叠谤补颈苍 functional activation and first mood episode in youth at risk for bipolar disorder.听锘� 听J Affect Disord. 2023;331:238-244. doi:
44.Suominen
听K锘�, Mantere
听O锘�, Valtonen
听H锘�,
听et al. 听Early age at onset of bipolar disorder is associated with more severe clinical features but delayed treatment seeking.听锘� 听Bipolar Disord. 2007;9(7):698-705. doi:
45.Parry
听P锘�, Allison
听S锘�, Bastiampillai
听T锘�. 听鈥楶ediatric bipolar disorder鈥� rates are still lower than claimed: a re-examination of eight epidemiological surveys used by an updated meta-analysis.听锘� 听Int J Bipolar Disord. 2021;9(1):21. doi:
46.Maxwell
听SE锘�, Lau
听MY锘�, Howard
听GS锘�. 听Is psychology suffering from a replication crisis? what does 鈥渇ailure to replicate鈥� really mean?听锘� 听Am Psychol. 2015;70(6):487-498. doi:
47.Tejavibulya
听L锘�, Rolison
听M锘�, Gao
听S锘�,
听et al. 听Predicting the future of neuroimaging predictive models in mental health.听锘� 听Mol Psychiatry. 2022;27(8):3129-3137. doi:
48.Marek
听S锘�, Tervo-Clemmens
听B锘�, Calabro
听FJ锘�,
听et al. 听Reproducible brain-wide association studies require thousands of individuals.听锘� 听狈补迟耻谤别. 2022;603(7902):654-660. doi:
49.National Institutes of Health. Race and natural origin. Accessed October 5, 2023.
50.McLaren
听DG锘�, Ries
听ML锘�, Xu
听G锘�, Johnson
听SC锘�. 听A generalized form of context-dependent psychophysiological interactions (gPPI): a comparison to standard approaches.听锘� 听狈别耻谤辞颈尘补驳别. 2012;61(4):1277-1286. doi:
51.Picard
听RR锘�, Cook
听RD锘�. 听Cross-validation of regression models.听锘� 听J Am Stat Assoc. 1984;79(387):575-583. doi:
52.Marangoni
听C锘�, De Chiara
听L锘�, Faedda
听GL锘�. 听Bipolar disorder and ADHD: comorbidity and diagnostic distinctions.听锘� 听Curr Psychiatry Rep. 2015;17(8):604. doi:
53.Benasi
听G锘�, Fava
听GA锘�, Guidi
听J锘�. 听Prodromal symptoms in depression: a systematic review.听锘� 听Psychother Psychosom. 2021;90(6):365-372. doi:
54.Chen
听MH锘�, Chen
听YS锘�, Hsu
听JW锘�,
听et al. 听Comorbidity of ADHD and subsequent bipolar disorder among adolescents and young adults with major depression: a nationwide longitudinal study.听锘� 听Bipolar Disord. 2015;17(3):315-322. doi:
55.Coxe
听S锘�, West
听SG锘�, Aiken
听LS锘�. 听The analysis of count data: a gentle introduction to poisson regression and its alternatives.听锘� 听J Pers Assess. 2009;91(2):121-136. doi:
56.Greene
听WH锘�. Accounting for excess zeros and sample selection in poisson and negative binomial regression models. Published online March 1, 1994. Accessed February 26, 2023.
57.Lambert
听D锘�. 听Zero-inflated poisson regression, with an application to defects in manufacturing.听锘� 听罢别肠丑苍辞尘别迟谤颈肠蝉. 1992;34(1):1-14. doi:
58.Morin
听EL锘�, Howell
听BR锘�, Feczko
听E锘�,
听et al. 听Developmental outcomes of early adverse care on amygdala functional connectivity in nonhuman primates.听锘� 听Dev Psychopathol. 2020;32(5):1579-1596. doi:
59.Pitkanen
听A锘�. Connectivity of the rat amygdaloid complex. the amygdala, a functional analysis. Published online 2000. Accessed March 13, 2023.
60.Huang
听TN锘�, Hsu
听TT锘�, Lin
听MH锘�,
听et al. 听Interhemispheric connectivity potentiates the basolateral amygdalae and regulates social interaction and memory.听锘� 听Cell Rep. 2019;29(1):34-48.e4. doi:
61.Irwin
听W锘�, Anderle
听MJ锘�, Abercrombie
听HC锘�, Schaefer
听SM锘�, Kalin
听NH锘�, Davidson
听RJ锘�. 听Amygdalar interhemispheric functional connectivity differs between the non-depressed and depressed human brain.听锘� 听狈别耻谤辞颈尘补驳别. 2004;21(2):674-686. doi:
62.Zald
听DH锘�, Donndelinger
听MJ锘�, Pardo
听JV锘�. 听Elucidating dynamic brain interactions with across-subjects correlational analyses of positron emission tomographic data: the functional connectivity of the amygdala and orbitofrontal cortex during olfactory tasks.听锘� 听J Cereb Blood Flow Metab. 1998;18(8):896-905. doi:
63.Zhou
听Y锘�, Wang
听Z锘�, Qin
听LD锘�,
听et al. 听Early altered resting-state functional connectivity predicts the severity of post-traumatic stress disorder symptoms in acutely traumatized subjects.听锘� 听PLoS One. 2012;7(10):e46833. doi:
64.Hrybouski
听S锘�, Aghamohammadi-Sereshki
听A锘�, Madan
听CR锘�,
听et al. 听Amygdala subnuclei response and connectivity during emotional processing.听锘� 听狈别耻谤辞颈尘补驳别. 2016;133:98-110. doi:
65.Milazzo
听AC锘�, Ng
听B锘�, Jiang
听H锘�,
听et al. 听Identification of mood-relevant brain connections using a continuous, subject-driven rumination paradigm.听锘� 听Cereb Cortex. 2016;26(3):933-942. doi:
66.Harmon-Jones
听E锘�, Abramson
听LY锘�, Sigelman
听J锘�, Bohlig
听A锘�, Hogan
听ME锘�, Harmon-Jones
听C锘�. 听Proneness to hypomania/mania symptoms or depression symptoms and asymmetrical frontal cortical responses to an anger-evoking event.听锘� 听J Pers Soc Psychol. 2002;82(4):610-618. doi:
67.Bertocci
听MA锘�, Afriyie-Agyemang
听Y锘�, Rozovsky
听R锘�,
听et al. 听Altered patterns of central executive, default mode and salience network activity and connectivity are associated with current and future depression risk in two independent young adult samples.听锘� 听Mol Psychiatry. 2023;28(3):1046-1056. doi:
68.Mulders
听PC锘�, van Eijndhoven
听PF锘�, Schene
听AH锘�, Beckmann
听CF锘�, Tendolkar
听I锘�. 听Resting-state functional connectivity in major depressive disorder: a review.听锘� 听Neurosci Biobehav Rev. 2015;56:330-344. doi:
69.Hare
听TA锘�, Tottenham
听N锘�, Davidson
听MC锘�, Glover
听GH锘�, Casey
听BJ锘�. 听Contributions of amygdala and striatal activity in emotion regulation.听锘� 听Biol Psychiatry. 2005;57(6):624-632. doi:
70.Fournier
听JC锘�, Bertocci
听M锘�, Ladouceur
听CD锘�,
听et al. 听Neural function during emotion regulation and future depressive symptoms in youth at risk for affective disorders.听锘� 听狈别耻谤辞辫蝉测肠丑辞辫丑补谤尘补肠辞濒辞驳测. 2021;46(7):1340-1347. doi:
71.Yoon
听S锘�, Kim
听TD锘�, Kim
听J锘�, Lyoo
听IK锘�. 听Altered functional activity in bipolar disorder: a comprehensive review from a large-scale network perspective.听锘� 听叠谤补颈苍 Behav. 2021;11(1):e01953. doi: