Introduction

Irritability, characterized by heightened susceptibility to frustration and anger, is a transdiagnostic symptom frequently noted as a primary reason for a child’s referral to mental health services [1,2,3]. Recent research has highlighted the complex interplay between neurostructural abnormalities and the manifestation of irritability in children. Specifically, irritability may be the result of disruptions in neurocircuits that contribute to attentional bias toward negative stimuli or the negative interpretation of neutral stimuli [3,4,5,6]. Others have implicated differences in sensory processing, with those with irritability showing heightened sensitivity to environmental stimuli, such as noise, light, or tactile sensations [7,8,9]. Alternatively, irritability may be due to differences in memory-related brain regions, which could lead to enhanced retrieval of negative associations and expectations of the unpleasant nature of encountering the perceived irritant [10, 11]. Finally, irritability could result from increased engagement of self-referential neurocircuitry, where the individual’s subjective experience is centered on personal distress, amplifying the perception of irritability [6, 12, 13]. Despite evidence of the importance of irritability symptoms in predicting concurrent and future mental health problems [3, 14, 15], there is a lack of research examining the neurobiological mechanisms underlying irritability in children, especially for brain structure. Additionally, work in this area has not determined whether these neural correlates of irritability are distinct from, or rather a component of, broader psychopathology. Thus, research is needed that examines neurostructural differences associated with irritability symptoms in younger samples.

Across multiple cross-sectional studies, differences in gray matter volume, cortical thickness, and surface area have been associated with irritability; however, there remain discrepancies in terms of the areas implicated. Studies focusing on gray matter volume show that irritability in youth is associated with smaller volumes in the putamen, internal capsule, frontal areas, and cingulate [16,17,18,19]. Other research instead demonstrates an association with larger volumes in the cingulate, in addition to larger hippocampal, insula, and medial orbitofrontal regions [19]. However, the sample in the Dennis et al. [19] study was unique in that they recruited youth specifically for exposure to early life stress. In terms of cortical thickness, higher endorsement of irritability has been associated with thinner cortices in temporal regions and the superior frontal gyrus [16, 20]. One study found that irritability was associated with less surface area in the bilateral rostral prefrontal cortex and the precuneus [21]. There are also studies demonstrating no significant relationship between irritability symptoms and regional gray matter volume, cortical thickness, or surface area [22,23,24]. Since these studies span large age ranges from childhood to adulthood, discrepant findings may be due to complex changes across development.

Longitudinal research further shows that irritability is associated with changes in brain structure over time. One study found that higher school-age irritability was associated with smaller gray matter volume in regions associated with emotion in pre-adolescence and with larger gray matter volume in areas such as the cerebellum [25]. Other studies show that higher irritability was associated with larger volumes and less volume contraction in brain regions that tend to decrease in volume over the developmental period [19]. In terms of cortical thickness, elevated irritability is associated with cortical thinning in brain regions implicated in emotion regulation [20]. Pagliaccio et al. identified distinct trajectories of irritability symptoms and found three classes of longitudinal irritability—elevated irritability that decreases over time, consistently low irritability, and consistently elevated irritability—with those exhibiting consistently elevated irritability showing a thicker cortex in the left superior frontal and temporal gyri and the right inferior parietal lobule [26]. Divergent longitudinal findings in the literature may be due to relatively small sample sizes and analyzing brain differences across a wide age range, highlighting the need for research focused on distinct periods of development.

While some neurostructural correlates of irritability have been identified, it remains unclear whether these associations are specific to irritability or indicative of mental health symptoms more generally. Research suggests that many mental health disorders, especially anxiety and depression, share common features such as negative affect or neuroticism [27, 28]. Additionally, a broad range of mental health conditions are characterized by similarities in brain activation patterns [29, 30] and impaired inhibitory control [31,32,33], particularly in regions associated with processing and assigning salience to environmental stimuli, such as the salience network [34]. The neurostructural and functional alterations observed across various mental health disorders highlight their central role in general psychopathology. Given that irritability is one component of negative affect and neuroticism [35, 36], the neurostructural substrates of irritability may simply reflect associations with broader mental health symptoms. A number of studies examining structural associations with negative affect/neuroticism have identified many regions that are similar to those found in the irritability literature [37, 38]. Thus, it is important to investigate whether these neurostructural differences are specific to irritability or reflect an underlying component of psychopathology more generally.

Most studies examining neural correlates of irritability in youth have relied on relatively modest sample sizes, have used wide age ranges which can obscure developmental changes, or have focused on adolescence specifically. Given that childhood irritability serves as a significant risk factor for developing future anxiety and depressive disorders [14], it is important to investigate neural substrates of irritability in children and determine whether structural differences are associated with later symptoms. Moreover, it is important to establish whether these neurostructural differences are unique to irritability or reflect a broader aspect of psychopathology. Therefore, the purpose of the current study was to examine the neural correlates of irritability in 9–10-year-old children with a focus on identifying structural differences that may be specific to irritability symptoms. To accomplish this, we utilized data from 11,868 children from the Adolescent Brain Cognitive Development (ABCD) Study to examine associations at 9–10 years of age and over the next 3 years [39]. Using confirmatory factor analysis, we derived a latent variable of irritability and related this to gray matter volume, cortical thickness, and surface area, both with and without controlling for anxiety symptoms, depressive symptoms, or total mental health symptoms. Given the divergent findings in the literature, we conducted an exploratory whole-brain analysis as we had no specific hypotheses regarding which brain regions would be associated with irritability.

Methods and materials

Participants

The present analyses used the baseline and 3rd-year follow-up data from release 5.1 of the ABCD Study [39]. The use of this de-identified dataset was approved by the institutional review board at Vanderbilt University. Participants in the ABCD Study were recruited at 21 sites across the United States [39, 40]. The baseline time point of the ABCD Study includes data from 11,868 children between 9 and 10 years of age. After excluding cases with missing data or who failed to pass quality assurance measures, the final sample size included in analyses was N = 10,647. A summary of the demographic characteristics of the sample used in analyses can be found in Table 1. For details on the generalizability of the ABCD Study sample, see the Supplementary Information.

Table 1 Demographic characteristics of the sample included in analyses of associations between irritability and brain structure (N = 10,647).
Full size table

Deriving a latent measure of irritability

Irritability was measured using the Child Behavior Checklist (CBCL) [41]. The CBCL is a parent-report measure that contains 119 items describing behaviors and emotions in children. Items were rated on a 3-point scale (0 = not true (as far as you know), 1 = somewhat or sometimes true, and 2 = very true or often true). Irritability was defined as a latent variable comprised of five items from the CBCL, as has been done previously [42]. Standardized loadings and a scree plot are shown in Fig. 1a, b. Of note, prior work on irritability has examined two approaches for defining irritability: five CBCL items or the five CBCL items plus three items from the Kiddie Schedule for Affective Disorders and Schizophrenia [42]. As both approaches showed comparable internal consistency, we chose to use the five CBCL items in the present study. These five items showed good internal consistency in the current study (Cronbach’s α = 0.80). There was very little missing data on this measure, with 0.001% of the sample missing values on these items. For additional details on the creation of the latent irritability factor, see the Supplementary Information.

Fig. 1: Creating a latent irritability variable.
figure 1

A We used confirmatory factor analysis to understand how five irritability items from the Child Behavior Checklist relate to each other. The standardized loadings (the coefficients that represent the strength of the relationship between observed variables and the underlying latent factor) are provided. The results showed that this model fits the data well. B The scree plot shows the number of possible factors on the x-axis and the eigenvalues (which indicate how much variance each factor explains) on the y-axis. The dashed line at an eigenvalue of 1 represents a standard cutoff point. If a factor’s eigenvalue is 1 or higher, it is generally retained in the model. The scree plot indicates a unidimensional measure of irritability.

Full size image

Additional covariate subscales

To examine anxiety and depressive symptoms as separate covariates, we used the DSM-oriented CBCL subscales: Anxiety Problems (6 items measuring anxiety symptoms) and Affective Problems (14 items measuring depressive symptoms) [43]. We also created a CBCL total score by summing all CBCL items except the five items used in the irritability latent factor.

Image acquisition, processing, and quality assurance

Image acquisition, processing, and quality assurance procedures have been previously documented [40, 44]. Additional details regarding imaging data exclusions can be found in previous work [44] and in the Supplementary Information.

Statistical analyses

Analyses were conducted using structural equation modeling in Mplus version 8.8. We used the mean-adjusted and variance-adjusted weighted least squares estimator and pairwise deletion for missing data. Confirmatory factor analysis was used to derive a latent measure of irritability. Using this latent factor, we examined associations between irritability, gray matter volume, cortical thickness, and surface area in 68 cortical regions (34 regions in each hemisphere) derived from a surface-based atlas procedure [45]. In addition, for gray matter volume we tested 19 subcortical regions derived by an automated labeling procedure [46]. To account for known issues with the ABCD Study data, all analyses (1) incorporated clustering by family membership to account for siblings and multiple births (twins and triplets), (2) stratified based on site to account for site differences, (3) weighted the data by the post-stratification weights provided by the ABCD Study to make the sample more representative of the US population in terms of demographics, (4) weighted the data by non-participation weights to adjust for differences between the included and excluded samples as our prior work has shown that those excluded for failure to meet neuroimaging quality control differ on demographic variables [47], and (5) covaried by scanner model to control for differences between scanners. For each brain region, we investigated associations between irritability and brain structure (gray matter volume, cortical thickness, or surface area), controlling for age, sex, race/ethnicity, MRI scanner model, parent’s highest level of education as a proxy for socioeconomic status, intracranial volume (ICV), and current medication usage. Additionally, to test the specificity of our findings, we included anxious symptoms, depressive symptoms, or total psychopathology symptoms as an additional covariate. We then tested whether brain structure at baseline predicted future irritability symptoms collected at the 3rd-year follow-up session. To control for multiple testing across regions, we used the false discovery rate (q < 0.05) correction from the stats package in R version 4.2.1 (http://www.r-project.org/).

Results

Irritability is associated with smaller brain volumes in temporal and parietal regions

Our results showed that the latent irritability factor was associated with smaller brain volumes in primarily temporal and parietal regions. Out of the 68 cortical and 19 subcortical regions examined, irritability was inversely associated with gray matter volume in 11 cortical regions, but no subcortical regions (Fig. 2; Supplementary Fig. 1; Supplementary Table 1). Some of these associations were bilateral, including the inferior parietal lobe, precentral gyrus, and postcentral gyrus. Other regions that were significantly associated with irritability included the left superior temporal lobe, left supramarginal lobe, right inferior temporal lobe, right middle temporal lobe, and right posterior cingulate gyrus. No other gray matter volume regions were significantly associated with the latent irritability factor. Next, we included anxiety symptoms, depressive symptoms, or total psychopathology symptoms as additional covariates. When anxious symptoms were included in our model, brain regions significantly associated with irritability included the left fusiform gyrus, bilateral inferior parietal lobe, bilateral precentral gyrus, left superior temporal lobe, left supramarginal lobe, right middle temporal lobe, right banks of the superior temporal sulcus, and right posterior cingulate gyrus (Supplementary Table 2). When depressive symptoms were added as a covariate, the left superior temporal gyrus, right inferior parietal gyrus, right middle temporal gyrus, and right precentral gyrus remained significantly associated with irritability (Supplementary Table 3). Finally, when we added total psychopathology symptoms as a covariate, the only region that remained significant was the right inferior parietal lobe (Supplementary Table 4). Effect sizes for these associations are shown as standardized beta estimates and R-squared values in all tables.

Fig. 2: Irritability is associated with smaller gray matter volumes in primarily temporal and parietal regions.
figure 2

Structural equation modeling that controlled for age, sex, race/ethnicity, scanner model, medication use, parental education, and total intracranial volume revealed that greater latent irritability scores were associated with smaller volumes in the bilateral inferior parietal lobes, bilateral precentral and postcentral gyri, left superior temporal lobe, left supramarginal lobe, right inferior temporal lobe, right middle temporal lobe, and right posterior cingulate gyrus. The significant cortical regions derived from a surface-based atlas procedure [45] are shown. We utilized the false discovery rate (q < 0.05) to account for multiple comparisons.

Full size image

Gray matter volume does not predict future irritability symptoms

Given the significant association between irritability and gray matter volume at baseline, we next examined whether smaller volumes predicted irritability symptoms at a future time point. We tested regional gray matter volumes as predictors of irritability at the 3rd-year follow-up, controlling for age, sex, race/ethnicity, scanner model, parent’s highest level of education, medication use, total ICV, and baseline irritability symptoms. The results showed no significant associations between gray matter volume at baseline and future irritability symptoms (Supplementary Table 5).

Lack of associations for cortical thickness and surface area

Finally, we repeated the above analyses for cortical thickness and surface area in the 68 cortical regions. There were no significant associations between irritability and cortical thickness (Supplementary Table 6) or surface area (Supplementary Table 7).

Discussion

The current study investigated the associations between irritability symptoms and gray matter volume, cortical thickness, and surface area while controlling for other psychopathology symptoms in a large sample of children. Additionally, this study examined whether baseline structural measures were predictive of future irritability symptoms. Irritability was associated with smaller volumes in primarily temporal and parietal regions at baseline, consistent with previous work [16,17,18, 48]. The analyses controlled for known confounds including age, sex, race/ethnicity, scanner model, parental education, medication use, and total ICV. Additionally, we controlled for anxious, depressive, and general psychopathology symptoms to test the specificity of our findings. While many regions continued to be associated with irritability after controlling for anxious and depressive symptoms, the results became predominantly nonsignificant when controlling for general psychopathology. This may not be surprising as irritability is a highly transdiagnostic symptom that is apparent in many mental health disorders [8, 49], making it difficult to disentangle the relative contribution of irritability to structural differences above and beyond general psychopathology.

In terms of the longitudinal analyses, gray matter volume did not predict irritability at the 3rd-year follow-up when controlling for baseline irritability. The lack of prediction at a future time point may suggest the presence of compensatory mechanisms, such as neuroplasticity or functional reorganization [50, 51], enabling children with smaller volumes in these regions to mitigate the risk of developing future irritability symptoms. Alternatively, it is possible that a longer duration between time points is necessary to observe an association between baseline brain volume and future psychopathology outcomes. It is also possible that irritability is not stable over time, reducing the likelihood of detecting an association between brain structure and future symptoms. The variability in brain development during this period, occurring at different rates in various brain regions, could account for the lack of predictive power for gray matter volume. It is important to distinguish between association and causation in this context, as observing a statistical association between gray matter volume and irritability does not necessarily imply that changes in volume cause changes in irritability. Numerous factors, including genetic predispositions, environmental exposures, and psychosocial contexts, contribute to brain development and behavioral outcomes [52,53,54,55]. Thus, these findings should not be interpreted as causal, as the observed associations between gray matter volume and irritability are likely due to a complex interplay of multiple contributing factors. While gray matter volume was not associated with future irritability in children in this sample, there could be other mechanisms or factors that influence the development of irritability over time. Our findings highlight the complexity of the relationship between brain structure and irritability symptoms and suggest that additional research is needed to fully understand these mechanisms.

No significant associations between irritability and cortical thickness or surface area were found. This contrasts with two studies linking higher irritability to thinner cortices in ages 8–22 years [16, 20] and one study showing irritability is associated with less surface area in a sample ranging from 5 to 21 years of age [21]. However, our lack of associations is consistent with two studies that also did not find significant relationships between irritability and cortical thickness or surface area [23, 24]. Notably, the studies that do find significant associations included broad age ranges extending into adulthood, whereas the current study focused on ages 9–10 years, raising the possibility that such associations may emerge later in development. Given that volume is the product of cortical thickness and surface area, it may be initially surprising that gray matter volume shows significant results while cortical thickness and surface area do not. However, prior work has shown that volume, cortical thickness, and surface area are distinct but related facets of brain structure, each offering unique information and showing different normative developmental trajectories [56, 57]. Future work collecting data from independent samples across various age groups, spanning childhood to adulthood, will be necessary to accumulate enough data to clarify the associations between irritability and structural brain development.

The regions associated with irritability in this study were primarily involved in emotional processing and social cognition, attention allocation, and movement/perception. The left superior temporal lobe, right inferior temporal lobe, right middle temporal lobe, and bilateral inferior parietal lobes were all associated with irritability in the current study. The temporal lobes are multifunctional and are associated with the processing of emotions, emotion regulation, and social cognition [58,59,60,61,62,63]. Structural differences in these regions may in part underlie the perception of negative social interactions in children with irritability, who tend to show a hostile attribution bias or the tendency to attribute hostile intent to others’ behavior [3]. This may be targeted in clinical interventions, with work showing promising results for altering negative social biases using cognitive bias modification programs, such as positive interpretation training [64]. Notably, we did not find significant differences in frontal regions commonly associated with emotional regulation (anterior cingulate and prefrontal cortices), in contrast to prior work on irritability in specific diagnostic groups [18, 65]. Given the young age of our sample, associations between irritability and frontal brain regions may not become apparent until later in adolescence. It is also possible that functional differences in these regions might exist despite the absence of structural variations [66].

Additionally, the right inferior parietal lobe was the only brain region that was significantly associated with irritability after controlling for total psychopathology symptoms, suggesting a more specific relationship between irritability and this region. The inferior parietal lobe is associated with attention allocation [62], which is consistent with research showing that youth with irritability symptoms show attentional bias to threatening or angry faces even while controlling for confounds such as anxiety levels [4, 67], suggesting that one possible mechanism for the manifestation of irritability may be a tendency to focus on negative stimuli. Thus, the finding of differences in the inferior parietal lobes in the current study supports prior literature demonstrating that these lobes are particularly relevant to the association between irritability and reward- and threat-related neural processes [68, 69]. Relatedly, we found smaller volumes in the supramarginal gyrus, a region implicated in interoception or sensing the internal state of the body, which may be related to attention reorientation [70]. Our results also showed that irritability was associated with the posterior cingulate gyrus, a central hub in the default mode network that is related to internally directed processes and regulating the focus of attention [71] and has been implicated in studies of irritability [72]. It is possible that greater sensitivity to one’s internal thoughts and physical states (e.g., heart rate increasing, body temperature rising) coupled with an attentional bias towards negative stimuli and a tendency to attribute hostile motives to others may increase the likelihood that an individual will report feeling irritable.

The final group of regions found in the current study are related to movement and perception and include the precentral and postcentral gyri [73]. Our results are consistent with prior work in youth with irritability symptoms showing smaller gray matter volumes in motor-related regions [17, 18]. Given that irritability is a transdiagnostic symptom associated with many forms of psychopathology [1, 73, 74] and that many disorders implicate the precentral and postcentral gyrus [65, 75,76,77,78], the finding of differences in movement-related regions may be expected. These results are also consistent with prior work showing that irritability and agitation (an unpleasant state of restlessness and hyperarousal that has links with motor regions) commonly present together [79].

Clinical implications

The transdiagnostic nature of irritability has important implications for diagnosis and treatment. Recognizing irritability as a transdiagnostic trait could potentially lead to more dimensional approaches in mental health assessment that consider irritability as a core feature across many disorders. This shift could enable tailored interventions targeting the underlying mechanisms shared by disorders [80], while also aiding in the early identification of individuals at risk for developing comorbid conditions, thus facilitating more effective prevention and intervention strategies. Existing interventions for irritability are mainly focused on symptom reduction and relapse prevention, with work demonstrating that approaches promoting resilience in both parents and children are promising for early prevention, not only for irritability but also for other comorbid symptoms as well [81]. However, randomized controlled trials in children with severe irritability are needed to assess the efficacy of potential pharmacological and psychological treatments.

Strengths and limitations

This study’s strengths include a large sample size and a narrowly defined age range, which enables us to examine the relationship between brain structure and irritability while avoiding age-related variance. While the ABCD Study sample is fairly representative in terms of demographic and socioeconomic distribution, it is not perfectly representative. Therefore, another strength of this study was the use of post-stratification weights to adjust the sample to be more representative of the US population demographics. Additionally, while previous studies focus on one point in time, we leverage longitudinal data to examine whether neurostructural measures predict future irritability symptoms.

Limitations of the current study include the use of parent-reported irritability symptoms which account for parent perceptions but do not reflect child self-report or clinician ratings. As previous research has demonstrated, parent and child-reported irritability symptoms tend to diverge [82], thus, future work should include measures of self-reported irritability as well. The ABCD Study will include a self-report version of the CBCL (the Youth Self-Report) when the participants turn 14 years old [83], which will allow future work to examine these associations using self-reported irritability symptoms. Additionally, we defined irritability as a unidimensional construct and did not differentiate phasic (i.e., outburst) versus tonic (i.e., mood) components of irritability in our analyses. Future work could aim to assess irritability as multifaceted, either through data-driven exploration of subfactors of irritability or using measures that are able to distinguish between these aspects of irritability [84, 85]. Another limitation of our study is the exclusive focus on neurobiological factors, without incorporating environmental and other influences that also play a critical role in the development and manifestation of irritability symptoms. Given the multidimensional and context-dependent nature of irritability, future work should incorporate a more integrative approach that includes genetic, environmental, and psychosocial factors to better understand the complex predictors of irritability. Finally, the effect sizes found in our analysis of brain structure and irritability were relatively small; however, previous studies with large samples have also consistently shown small but reliable associations between brain structure and behavior [86].

Conclusions

The current study expands on prior work by examining the neurostructural correlates of irritability in youth by using a large, well-defined sample, creating a latent measure of irritability, and leveraging a longitudinal design to predict future symptoms. Our findings underscore the importance of brain regions associated with emotional processing, social cognition, attention allocation, and movement/perception in understanding irritability symptoms in children. These results support theories positing deficits in socioemotional and attentional neural circuitry as key features of irritability and demonstrate that these neurostructural differences are apparent at 9–10 years of age. Given that structural differences in these areas are implicated in many mental health problems, the current study supports the conceptualization of irritability as a transdiagnostic trait across disorders. Taken together, the results of this study provide a deeper understanding of neurostructural substrates that may be related to multiple disorders that involve irritability as a core symptom, thereby increasing our understanding of the neurobiological mechanisms underlying diverse mental health conditions.