Neurostructural Differences Associated With Prodromal Mania Symptoms in Children

Authors: Camille Archer, Amy Milewski, Hee Jung Jeong, Gabrielle E. Reimann, E. Leighton Durham, Antonia N. Kaczkurkin

Abstract

Introduction

Prodromal symptoms of mania in children are predictive of the later development of bipolar disorder; yet, the neurostructural correlates of these early symptoms remain poorly understood. This study aimed to investigate the association between prodromal mania symptoms and brain structure in a large cohort of children.

Methods

We analyzed data from 10,662 nine‐ to 10‐year‐old children from the Adolescent Brain Cognitive Development (ABCD) Study, employing structural equation modeling to examine the concurrent and longitudinal associations between prodromal mania symptoms and cortical and subcortical gray matter volume.

Results

After adjusting for multiple comparisons and controlling for age, sex, scanner model, socioeconomic status, and medication use, we found that baseline mania symptoms were associated with reduced gray matter volume across both cortical and subcortical areas, suggesting a global effect. These findings were further supported by the loss of these effects when total intracranial volume was included as an additional covariate, suggesting that smaller overall brain size, rather than specific regional effects, is related to prodromal mania symptoms. Lastly, longitudinal analyses revealed that brain volume at baseline did not predict prodromal mania symptoms at the second‐year follow‐up.

Conclusion

Our results support the structural differences observed in adults with bipolar disorder in prior work and refine our understanding of the neurostructural correlates of prodromal mania symptoms in children. These findings could enhance early identification and intervention efforts for youth at risk of developing bipolar disorder.

Introduction

Bipolar disorder is a severe mental health condition marked by alternating periods of mania and depression and often leads to an increased risk of suicide, as well as cognitive and functional impairments that worsen over time (Lan et al. 2014; Maletic and Raison 2014; Muneer 2016). Although the mean age of onset for the first manic, hypomanic, or depressive episode is around 18 years for bipolar I disorder and in the mid‐20s for bipolar II disorder (American Psychiatric Association 2022), bipolar spectrum disorders (including bipolar I, bipolar II, cyclothymic disorder, and other specified or unspecified bipolar disorders) have also been observed in children and adolescents (Goldstein et al. 2017). Diagnosing bipolar spectrum disorders in youth is challenging, often resulting in children and adolescents receiving a diagnosis of other specified bipolar and related disorder (Goldstein et al. 2017). Nevertheless, a significant portion of children and adolescents initially diagnosed with other specified bipolar and related disorder eventually develop bipolar I or II, suggesting that prodromal mania symptoms may serve as early indicators of the disorder (Axelson et al. 2011; Conroy et al. 2018). Prodromal mania symptoms, such as excitation, mood swings, irritability, anxiety, hyperactivity, and sleep disturbances, can manifest months or years before the first full manic or hypomanic episode, which underscores the importance of early detection and intervention (Correll et al. 2014; De Pablo et al. 2020; Skjelstad et al. 2010). Despite growing awareness of the importance of detecting prodromal mania symptoms in youth, there is a lack of research on the neurostructural correlates of these early symptoms.

Neurostructural differences are evident in individuals with bipolar disorder throughout the lifespan. In adults with bipolar disorder, research consistently shows cortical alterations in frontal, temporal, and parietal brain regions (Abé et al. 2016; Hibar et al. 2018). Specifically, adults with bipolar disorder present with smaller gray matter volume in the dorsomedial and ventromedial prefrontal cortex, anterior cingulate cortex, bilateral insula, and superior temporal gyrus (Wang et al. 2019; Wise et al. 2017). Additionally, smaller gray matter volumes in the amygdala and hippocampus are related to bipolar disorder in adult populations (Angelescu et al. 2021; Hibar et al. 2016). In contrast, adults with bipolar disorder show greater gray matter volume in areas such as the inferior temporal gyrus and bilateral middle frontal gyrus, as well as cerebellar regions (Wise et al. 2017).

When examining structural differences in youth with bipolar disorder, high‐risk children and adolescents show smaller cortical gray matter volume in frontal regions, including the inferior frontal gyrus, lateral orbitofrontal cortex, frontal pole, and rostral middle frontal gyrus (Lim et al. 2013; Roberts et al. 2022). Bilateral amygdala and limbic region volumes are also smaller in adolescents with bipolar disorder, with longitudinal studies showing that amygdala volume reduction is specific to youth compared to adults (Förster et al. 2023; Long et al. 2023). In contrast, some work has found that youth with bipolar disorder show larger volumes in the right medial orbitofrontal cortex and left superior frontal gyrus (Long et al. 2023; Roberts et al. 2022). Additionally, children with bipolar disorder have been shown to have overall smaller total cerebral brain volume in some studies (Frazier, Ahn, et al. 2005).

Taken together, the existing literature highlights both distinct and shared neurostructural differences in adults and children with bipolar disorder. However, much of the research on structural differences in youth has focused on pediatric or adolescent samples with a diagnosis of a bipolar spectrum disorder (Dickstein et al. 2005; Frazier, Breeze, et al. 2005; Gogtay et al. 2007; Kaur et al. 2005; Toma et al. 2019). Given that subthreshold manic symptoms are often present for a lengthy period of time before the first full manic or hypomanic episode (Correll et al. 2014; Van Meter et al. 2021), studies are needed that examine the neurostructural correlates of prodromal mania symptoms in children prior to receiving a bipolar disorder diagnosis. Furthermore, many of the studies on structural differences in children with bipolar disorder have relied on relatively small samples sizes with varying age ranges. Research involving a large cohort of children with a narrow age range would be useful to determine whether neural correlates of prodromal mania symptoms can be detected before the onset of the first manic or hypomanic episode.

The current study aimed to overcome these previous limitations by exploring the relationship between a dimensional measure of prodromal mania symptoms and regional cortical and subcortical brain volumes in children. We examined this association in a large, community‐based sample of 9‐ to 10‐year‐olds (N = 10,662) at baseline and again over a 2‐year period. We hypothesized that higher levels of prodromal mania symptoms would be associated with smaller brain volume in areas previously implicated in bipolar disorder, including frontal regions and the amygdala. Additionally, we predicted that smaller brain volume in these regions at baseline would predict higher levels of prodromal mania symptoms at the second‐year follow‐up.

Methods and Materials

Participants

This study drew on baseline and second‐year follow‐up data from the Adolescent Brain Cognitive Development (ABCD) Study, release 5.1 (Volkow et al. 2018). Recruitment for the ABCD Study took place in the United States across 21 sites, with the baseline sample size consisting of data from 11,868 children aged 9–10 years old (Casey et al. 2018; Volkow et al. 2018). The current analyses utilized data from 10,662 participants after excluding those with missing data or who failed to pass quality assurance measures. Table 1 provides further details on the final sample's demographic information. Vanderbilt University's Institutional Review Board approved the use of this publicly available, de‐identified dataset. Caregivers in the ABCD Study provided informed consent and minors provided informed assent.

Measures

Prodromal Mania Measure

The Parent General Behavior Inventory, 10‐Item Mania scale (PGBI‐10 M) was developed from the original 73‐item PGBI and includes the 10 most predictive items for assessing prodromal bipolar disorder symptoms, including manic and biphasic (mixed state including aspects of both mania and depression) symptoms (Freeman et al. 2012; Youngstrom et al. 2008). This caregiver‐rated measure has been shown to distinguish prodromal bipolar disorder symptoms from comorbid conditions such as unipolar depression and attention‐deficit/hyperactivity disorder (ADHD) (Barch et al. 2018; Youngstrom et al. 2008). The PGBI‐10 M shows excellent internal consistency (α = 0.92) and high correlations with earlier versions in prior work (Youngstrom et al. 2008).

Image Acquisition, Quality Assurance, and Processing

The ABCD Study collected MRI data on multiple models of 3 tesla (3T) General Electric Discovery MR750, Siemens Prisma, Siemens Prisma Fit, Philips Achieva dStream, and Philips Ingenia (Casey et al. 2018). Three‐dimensional T1‐ and T2‐weighted images of brain structure were collected, with whole brain T1‐weighted images obtained using the following TR (repetition time) 2400–2500 ms; TE (echo time) 2–2.9 ms; FOV (field of view) 256 × 240 to 256; FOV phase of 93.75%–100%; matrix 256 × 256; 176–225 slices; TI (inversion delay) 1060 ms; flip angle of 8°; and voxel resolution of 1 × 1 × 1 mm; total acquisition time was 7 min and 12 s for Siemens Prisma, 6 min and 9 s for GE 750, and 5 min and 38 s for Philips.

The imaging data were processed and analyzed by the ABCD Data Analysis and Informatics Center (DAIC) using the Multi‐Modal Processing Stream (MMPS), a software package created at the Center for Multimodal Imaging and Genetics (CMIG) at the University of California, San Diego (UCSD). Data underwent correction for gradient nonlinearity distortions, intensity scaling and homogeneity correction, registration to an averaged reference brain in standard space, and manual quality control. Using automated, atlas‐based, segmentation processes in FreeSurfer v.5.3, data underwent cortical surface reconstruction and subcortical segmentation. To quantify morphometric measures, DAIC used the average cortical thickness and volume in 68 cortical parcels of the Desikan–Killiany atlas (Desikan et al. 2006) and the average volume in 19 subcortical regions from the FreeSurfer subcortical atlas (Fischl et al. 2002). Trained technicians completed manual quality control on the images for motion, intensity homogeneity, white matter underestimation, pial overestimation, and magnetic susceptibility artifact. Additional details regarding the ABCD Study imaging procedures have been previously documented (Casey et al. 2018; Hagler et al. 2019).

Data Analysis

Utilizing structural equation modeling in Mplus version 8.8., brain volume was related to a dimensional measure of prodromal mania symptoms while controlling for age, sex, scanner model, socioeconomic status (SES; defined as parent's highest level of education), and medication use. For longitudinal analyses, baseline prodromal mania symptoms were added as a covariate. All analyses clustered based on family to account for twins and siblings and stratified based on site. Poststratification weights provided by the ABCD Study were also applied to account for discrepancies between the sample and the US population on key demographics, such as sex and race/ethnicity. As we have done in our previous work, nonparticipation weights were used to make the included and excluded samples more similar, given that participants who were included in the MRI subsample differ significantly on important demographics from those who were excluded for MRI data quality issues (Durham et al. 2021). The false discovery rate (FDR) was used to adjust p‐values to account for multiple comparisons.

Sensitivity Analyses

To further investigate the robustness of the primary findings, total intracranial volume was added as an additional covariate. This allows us to account for potential associations that may exist between brain volume and overall variations in cranium size. By including total intracranial volume in our analyses, we aimed to determine whether any regional volume effects exist, above and beyond global differences in brain size.

Data and Code Availability

The National Institute of Mental Health Data Archive (https://nda.nih.gov/abcd) provides access to the ABCD Study data. To access the current analyses’ scripts with a detailed description, visit https://github.com/VU‐BRAINS‐lab/Archer_Mania_Volume.

Results

Prodromal Mania Is Associated With Differences in Brain Structure at Baseline

Following FDR correction for multiple comparisons and controlling for age, sex, differences in scanner model, SES, and medication use, greater prodromal mania symptoms were associated with smaller cortical and subcortical gray matter volumes in all brain regions examined (p fdr‐values ≤0.05) (Table 2). Of these regions, several demonstrated a strong association with prodromal mania symptoms, including the left fusiform gyrus, bilateral precentral gyrus, right middle temporal gyrus, and bilateral thalamus (see Figure 1).

Total Intracranial Volume Sensitivity Analyses Reveal a Global Effect

Sensitivity analyses were conducted with total intracranial volume included as an additional covariate, alongside age, sex, scanner model, SES, and medication use. Although the main analyses showed an association between higher prodromal mania scores and smaller cortical and subcortical volumes at baseline, this relationship was no longer significant after controlling for intracranial volume. The inclusion of total intracranial volume resulted in the disappearance of significant effects in all brain regions tested (p fdr‐values ≥0.07) (Table 3).

No Evidence of Volume Predicting Future Mania Symptoms

Finally, we examined whether regional volume at baseline predicted future mania symptoms at the second‐year follow‐up. Analyses revealed that none of the regions examined showed a significant association with future prodromal mania symptoms. While some regions did show an association at uncorrected levels (Table 4), no brain regions were significantly associated with prodromal mania symptoms at the 2‐year follow‐up following FDR correction for multiple comparisons.

Discussion

We examined associations between prodromal mania symptoms and brain volume in a large sample of children and found that the neurostructural differences associated with these symptoms are apparent in late childhood. After adjusting for age, sex, SES, scanner model, and medication use and applying correction for multiple comparisons, higher mania symptoms were associated with smaller volume across all regions at baseline. However, these relationships no longer held once total intracranial volume was taken into account, suggesting these associations may reflect global differences in overall brain size rather than region‐specific alterations. Furthermore, baseline volume did not predict mania symptoms at the second‐year follow‐up. Together, these results suggest that while early prodromal mania symptoms are related to structural brain differences, the effects appear nonspecific to individual regions and may be largely attributable to overall brain volume.

Our findings are broadly consistent with prior research documenting smaller gray matter volumes in youth with bipolar disorder and in those at elevated risk. Similar to studies reporting reduced cortical volume in frontal regions (Lim et al. 2013; Roberts et al. 2022) and smaller amygdala volumes in youth with bipolar disorder (Förster et al. 2023; Long et al. 2023), we also observed that higher prodromal mania symptoms were associated with smaller regional volumes across multiple areas. Although most prior studies have relied on clinical or high‐risk samples, our results extend this literature by showing comparable directions of associations in a large, community‐based cohort of children who had not yet received a diagnosis. This suggests that structural correlates of mania symptoms may be detectable even before the onset of the first manic or hypomanic episode, supporting the idea that such neurobiological differences emerge early in development and may represent dimensional vulnerability markers along the bipolar spectrum.

Notably, the associations between mania symptoms and smaller regional volumes did not persist after adjusting for total intracranial volume, indicating that the effects we observed were global rather than localized to specific cortical or subcortical regions. This is consistent with prior work showing that children with bipolar disorder may have smaller overall cerebral volume (Frazier, Ahn, et al. 2005), suggesting that brain‐wide structural differences may be important predictors of symptom expression during development. One interpretation is that mania‐related neurostructural alterations in childhood reflect a general delay or difference in neurodevelopmental processes, such as synaptic pruning, cortical maturation, or global brain growth, rather than disruptions of particular circuits. However, it is also important to note that these associations may not be unique to mania, as smaller global brain volumes have also been linked to general psychopathology across multiple disorders (Durham et al. 2023, 2021; Kaczkurkin et al. 2020, 2019). Thus, this global pattern could suggest that prodromal mania symptoms in childhood are tapping into broad neurostructural differences evident across mental health conditions, which highlights the importance of considering global brain growth trajectories in models of early psychopathology.

Contrary to our hypotheses, baseline brain volume did not predict mania symptoms at the 2‐year follow‐up. This null result suggests that structural differences at ages 9 and 10 years may not be robust predictors of symptom trajectories over short developmental windows. One possibility is that structural alterations observed at this early age reflect transient maturational variability rather than stable risk markers. Another possibility is that brain–behavior relationships become more specific and predictive closer to the age of peak risk for manic episodes, which typically occurs in late adolescence or early adulthood (American Psychiatric Association 2022; Goldstein et al. 2017). Alternatively, symptom progression may be more strongly shaped by dynamic factors such as functional brain changes, environmental stressors, or gene–environment interactions, which could interact with structural differences later in development. The ongoing longitudinal follow‐up of the ABCD Study cohort into adolescence and early adulthood will be critical for determining whether neurostructural markers acquire greater predictive utility as prodromal symptoms consolidate and clinical episodes emerge.

This study offers several notable strengths. The ABCD Study's large, community‐based sample provides substantial statistical power to detect subtle effects and captures a wide spectrum of mania symptoms beyond clinical thresholds. By avoiding a narrow focus on children with the most severe symptoms, we reduced the risk of inflated effect sizes that can occur in case–control designs. The restricted age range of the sample also minimized the confounding effects of normative brain development. Together, these features enhance the generalizability of our findings to the broader population. At the same time, certain limitations should be acknowledged. Mania symptoms were assessed exclusively via caregiver report, which may introduce bias or reduce sensitivity to subtle symptom expression. In addition, while our effects were small, this pattern is common in large‐scale neuroimaging studies of brain–behavior relationships (Paulus and Thompson 2019). Small but reliable effects may nonetheless be informative at the population level, though their clinical utility remains to be determined.

This study provides a valuable step forward in understanding the link between mania symptoms and structural differences during late childhood. However, further work should be done to track the relationship between mania symptoms and brain structure from late childhood onward to better understand when this association stabilizes and/or diverges. Future work could also supplement the current findings by investigating other early indicators of bipolar disorder, such as inflammation. Prior work has implicated several hematological indices (including neutrophil count, platelet count, mean platelet volume, neutrophil‐to‐lymphocyte ratio, platelet‐to‐lymphocyte ratio, and monocyte‐to‐lymphocyte ratio) as potential predictors of mood state, with platelet‐to‐lymphocyte ratio in particular suggested as an independent predictor of mania and hypomania (Fusar‐Poli et al. 2021). Given the growing evidence that neuroinflammatory processes may influence brain structure and function, incorporating such markers into longitudinal neuroimaging studies could clarify potential mechanistic links between inflammation and brain changes associated with mania risk.

An additional important avenue for future research involves the integration of detailed treatment‐related variables into longitudinal studies of brain development in youth with prodromal mania symptoms. Although our current analyses controlled for baseline medication use, we did not have the ability to assess the nuanced clinical factors of treatment response or adherence over time. Prior work in mood disorders has shown that illness severity and treatment side effects are significant predictors of nonadherence, which in turn can influence clinical outcomes (Pompili et al. 2009). These issues are also likely to be highly relevant for youth at risk for bipolar disorder, as early treatment initiation and adherence may interact with neurodevelopmental processes. For example, inadequate treatment adherence during critical developmental windows could exacerbate mood symptoms and potentially contribute to neuroanatomical differences through mechanisms such as heightened allostatic load, recurrent mood episodes, or prolonged exposure to mood instability. Conversely, effective and sustained treatment engagement may help normalize developmental brain trajectories or mitigate progressive structural changes. Future studies that combine longitudinal neuroimaging with detailed treatment histories, adherence measures, and side effect profiles could clarify the extent to which these clinical variables moderate the relationship between prodromal mania symptoms and brain structure.

In summary, this study examined the neuroanatomical correlates of prodromal mania symptoms in a large sample of children, revealing that higher baseline mania symptoms were associated with smaller global gray matter volumes across cortical and subcortical regions. These associations did not persist when total intracranial volume was included as a covariate, further underscoring the importance of considering global brain metrics in studies of mood symptomatology. Taken together, our findings point to smaller brain size as a potential neurodevelopmental marker of early mania risk, emphasizing the need for longitudinal investigations that can disentangle causal pathways, identify early intervention targets, and ultimately improve outcomes for youth at heightened risk for bipolar disorder.

Author Contributions

Camille Archer: conceptualization, formal analysis, visualization, writing – original draft, writing – review and editing. Amy Milewski: conceptualization, formal analysis, visualization, writing – original draft, writing – review and editing. Hee Jung Jeong: writing – review and editing. Gabrielle E. Reimann: writing – review and editing. E. Leighton Durham: writing – review and editing. Antonia N. Kaczkurkin: conceptualization, formal analysis, funding acquisition, methodology, resources, supervision, visualization, writing – original draft, writing – review and editing.

Conflicts of Interest

The authors declare no conflicts of interest.

Ethics Statement

Vanderbilt University's Institutional Review Board approved the use of this publicly available, de‐identified dataset.

Consent

Caregivers in the ABCD Study provided informed consent and minors provided informed assent.

Peer Review

The peer review history for this article is available at https://publons.com/publon/10.1002/brb3.70894.