Does the Relationship between Age and Brain Structure Differ in Youth with Conduct Disorder?

Conduct disorder (CD) is a psychiatric disorder that usually emerges in childhood or adolescence. It is characterised by persistent antisocial behaviour that violates others’ rights or societal norms such as theft, physical violence towards people or animals, or vandalism (American Psychiatric Association [APA], 2013). The prevalence of CD is estimated at approximately 3–4% for boys and 1–2% for girls, and it is one of the leading reasons for referral to child mental health services in the UK and the USA (Fairchild et al., 2019).

CD is related to poorer physical health, educational problems, and an increased risk of mental and substance use disorders (Erskine et al., 2016; Rivenbark et al., 2018). Furthermore, while CD can be limited to childhood and/or adolescence, it can also persist across the lifespan and lead to the development of antisocial personality disorder and criminal behaviour in adulthood, thereby causing a high individual, economic and societal burden (Fairchild et al., 2019; Scott et al., 2001). Despite an increase in research on CD in recent years, it remains largely under-recognised and under-treated by mental health services (Coghill, 2013). Hence, developing new prevention and intervention models is imperative, meaning that research into the underlying causes and development of CD is crucial.

Neurodevelopmental theories of CD suggest that the condition is related to dysfunction in cortical and subcortical regions of the brain, thereby leading to neurocognitive deficits (Blair et al., 2014). Previous studies have identified impairments in decision making (Fairchild et al., 2009), emotion recognition, learning and regulation (Kohls et al., 2020; Raschle et al., 2019), as well as responses to fear-inducing stimuli (Fanti et al., 2016) and affective empathy (Martin-Key et al., 2020) in CD. Structural MRI studies have demonstrated alterations in cortical and subcortical brain regions associated with these neurocognitive functions in youth with CD when compared to healthy controls (HCs). For instance, two meta-analyses employing voxel-based morphometry (VBM) reported lower gray matter volume (GMV) in several brain regions including the insula, left fusiform gyrus, medial and superior frontal gyrus, as well as the amygdala, in the CD group (Noordermeer et al., 2016; Rogers & De Brito, 2016).

Other investigations exploring brain structural changes in individuals with CD have utilized surface-based morphometry (SBM). Unlike VBM, which estimates gray matter volume and density throughout the brain, SBM distinguishes between two aspects of cortical structure: cortical thickness (CT) and surface area (SA). Since CT and SA may have different developmental trajectories and genetic underpinnings (Panizzon et al., 2009), separate investigation of CT and SA provides valuable insights into the specific cortical changes associated with CD, enhancing spatial resolution and anatomical precision (Raznahan et al., 2011).

Lower CT has been identified in regions such as the ventromedial prefrontal cortex (vmPFC)/orbitofrontal cortex (OFC), fusiform and precentral gyrus, superior temporal cortex and precuneus in youth with CD (e.g., Fairchild et al., 2015; Smaragdi et al., 2017; Wallace et al., 2014). Additionally, reductions in SA were found in areas such as the precentral and inferior temporal cortex as well as the OFC (e.g., Fairchild et al., 2015; Jiang et al., 2015). These structural alterations have been linked to neurological functions such as emotion regulation, affective empathy and reward processing (Fairchild et al., 2015; Kohls et al., 2020). Overall, findings suggest that CD may be related to structural differences in cortical and subcortical regions that are implicated in neurocognitive processes that are impaired in youth with this condition.

While existing cross-sectional research has provided indirect support for the neurodevelopmental theory of CD, findings across the literature are still inconsistent. For example, reduced amygdala volume has been repeatedly reported in CD (Fairchild et al., 2011, 2013; Huebner et al., 2008; Sterzer et al., 2007) while other studies observed no significant group differences in this region (Dalwani et al., 2011; De Brito et al., 2009) – although it should be noted that the latter studies focused on youth with conduct problems rather than CD.

These replicability issues could be due to methodological differences between studies, small sample sizes, or a failure to control for influential factors such as sex (Smaragdi et al., 2017), callous-unemotional traits (Wallace et al., 2014), age-of-onset subtype (Jiang et al., 2015), and comorbid disorders such as attention-deficit/hyperactivity disorder (ADHD; Noordermeer et al., 2016). Additionally, heterogeneity of findings across studies may also be linked to differences in the age ranges that were included (e.g., Fairchild et al., 2015 [16–21 years]; Wallace et al., 2014 [10–18 years]; De Brito et al., 2009 [11–13 years]).

Neuroimaging studies of brain maturation in typically developing children suggest that CT, SA, as well as cortical and subcortical volumes follow both linear and non-linear patterns that depend on the specific brain region and parameter under investigation. Bethlehem et al. (2022) found CT to reach its peak around 1.7 years, followed by GMV at around 5.9 years and SA between 11–12 years, with regional variation (primary sensory regions maturing before those responsible for higher-cognitive functions), as well as sex differences (earlier brain maturation in females).

Regarding subcortical volumes, the caudate, putamen, globus pallidus, and nucleus accumbens reach their peaks during the first decade of life and show a gradual decrease thereafter. In contrast, developmental trajectories of the amygdala, thalamus, and hippocampus follow a more flattened inverted U-curve, with their peak volumes occurring later during the second and third decades of life, remaining relatively stable thereafter (Dima et al., 2022).

Given what is known about the brain’s protracted development, some studies of childhood psychiatric disorders have suggested that differences between clinical groups and HCs are not characterised by stable structural differences, but rather by deviations in maturational trajectories. For instance, Shaw and colleagues (2007) investigated cortical maturation in children with ADHD and showed that this group reached peak CT values approximately 3 years after their typically-developing peers (10.5 versus 7.5 years). Hence, it was proposed that ADHD is characterised by delayed brain maturation rather than stable differences in brain structure across the lifespan. Similarly, a large longitudinal MRI study investigating brain maturation in autism spectrum disorder (ASD) reported abnormally accelerated brain growth in early childhood (2–4 years) in this population. Subsequently, brain maturation appeared to slow down and was followed by accelerated decreases in total brain volume in the ASD group compared to HCs (Courchesne et al., 2011). These studies provide evidence for abnormal brain maturation in youth with neurodevelopmental disorders. In addition, they suggest that structural differences between cases and controls might be age-specific, which could result in inconsistent findings when investigating group averages in samples of varying age ranges, as has been the case in the CD literature.

Based on the previously described evidence of brain structural alterations in CD as well as findings of abnormal brain maturation in other neurodevelopmental disorders, it has been suggested that CD might also be associated with delays in brain development. De Brito and colleagues (2009) addressed this by comparing cross-sectional sMRI data of 11- to 13-year-old HCs to boys high in conduct problems and callous-unemotional traits (CP + CU). HCs displayed a decrease in gray matter concentration in the medial orbitofrontal and dorsal anterior cingulate cortex, whereas boys with CP+CU displayed little change or even an increase in gray matter concentration with age. Furthermore, Oostermeijer et al. (2016) performed a longitudinal study investigating the relationship between conduct problems and brain structure trajectories in adolescents (who were assessed at ages 12, 16 and 19). The sample was grouped based on conduct problems trajectories (stable low, intermediate and desisting), and analyses revealed slower cortical thinning in the desisting (i.e., high and then declining) CPs group in the anterior cingulate and dorsolateral prefrontal cortices. The desisting group also exhibited a sustained increase in hippocampal volume over time, whereas the intermediate group displayed minimal changes, and the stable low CP group demonstrated a comparatively modest and gradual increase. A recent longitudinal study by Albaugh and colleagues (2023) investigated the relationship between conduct problems and cortical thinning in a large sample (N = 1039) that were scanned twice over a period of 5 years from late adolescence. Using vertex-based models they found an association between higher levels of CP and accelerated cortical thinning in the bilateral insula and inferior parietal cortices, and the left inferior frontal gyrus, and rostral anterior and posterior cingulate cortices. These studies provide evidence for altered brain maturation in youth with conduct problems, however, they used non-clinical samples. Therefore, in order to gain further insight into the neurodevelopmental processes underlying CD, all participants in the current study underwent a comprehensive psychiatric assessment to confirm research diagnoses of CD in the clinical group, as well as assess for other common comorbid conditions. Additionally, we performed group-by-age (and group-by-age-by-sex) interaction analyses in a sample including a greater number of youth with CD than most previous analyses and further differentiated between different cortical structure parameters to provide a more comprehensive characterisation of cortical structure alterations.

Building on existing research, the current study aimed to enhance our understanding of neurodevelopmental processes in CD by investigating how brain development in youth with diagnoses of CD compares to that observed in HCs. The sample was selected from the largest cross-sectional structural MRI dataset of youth with CD available to date, and data from youth with CD and HCs were analysed to test for group differences in CT, SA and subcortical volumes and interactions between age and group as well as group-by-age-by-sex interactions. We aimed to replicate previous findings of lower CT or SA in frontal regions and lower volume of subcortical regions such as the amygdala in the CD group compared to HCs. Additionally, based on previous research with other childhood psychiatric disorders such as ADHD, we expected brain maturation to be delayed in CD compared to HC participants, particularly in frontal, temporal and limbic regions.

Table 1 reports the sample’s demographic and clinical characteristics. The CD and HC groups did not significantly differ in age, pubertal stage and handedness. However, female participants were significantly older and had higher pubertal development scores than male participants. Additionally, a significant interaction between sex and group was found for handedness, where males with CD were more likely to be left-handed than male HCs, while the opposite pattern was observed in females. As expected, the male and female CD groups had significantly more CD symptoms and higher levels of CU traits, but lower IQs than male and female HCs. Males with CD had higher rates of comorbid ADHD than their female counterparts. Unsurprisingly, the CD group had significantly higher rates of comorbid lifetime diagnoses of ODD, depression and anxiety disorders than the control group as the presence of psychiatric conditions was an exclusion criterion for the HCs. Furthermore, females with CD were significantly more likely to comorbid depression or anxiety disorders than their male counterparts.

The aim of this study was to investigate whether brain maturation is altered in youth with CD in comparison to healthy controls, using cross-sectional MRI data from a large sample spanning the developmental period from 9 to 18 years of age. Overall, the CD group showed lower surface area in a range of cortical regions including the frontal, temporal and parietal lobes (9 out of the 34 regions tested) as well as total surface area. The largest group differences were observed in the middle temporal and entorhinal cortex (albeit with relatively small effect sizes of d = 0.28). These results provide evidence that CD is associated with lower SA across a range of cortical regions. Notably, in addition to these main effects of group, we were also able to replicate the main effects of age that have been observed in previous normative studies (e.g., Raznahan et al., 2011) - CT values declined with age across the majority of brain regions studied, while SA and subcortical volume decreased in 40–50% of the investigated brain areas. The lack of significant interactions between group and age suggest that this downward trajectory did not differ between youth with CD and controls.

Although we hypothesized that brain maturation might be delayed in youth with CD, no significant interactions between group and age emerged that would have supported this prediction. This runs counter to previous studies that provided some evidence for altered neurodevelopmental trajectories related to conduct problems (Albaugh et al., 2023; De Brito et al., 2009; Oostermeijer et al., 2016). Notably, Albaugh and colleagues (2023) identified significant conduct problems-by-age interactions which indicated that conduct problems were associated with cortical thinning in the oldest age groups in their sample (18+ years). This may suggest that the lack of significant group-by-age interactions (for CT) in the present study might relate to the fact that we only included participants up to age 18 and that there were comparatively fewer ‘older’ participants. Furthermore, the brain undergoes extensive development before the age of 9, with cortical thickness peaking at around age 2 and surface area peaking at age 11 (Bethlehem et al., 2022), meaning if we had included data from participants both younger than 9 and older than 18, we may have found evidence for group-by-age interactions.

Alternatively, the lack of significant group-by-age interactions might also be due to Type II error, meaning that true, but small, effects are present but were not detected by the analyses. Recent research indicates that large sample sizes may be required to detect the small neuroanatomical alterations that we might expect in psychopathologies (Marek et al., 2022) and even larger samples are likely to be needed to perform well-powered interaction analyses. Therefore, while this study included a larger sample than many previous studies, our statistical power may still not have been sufficient to detect the small interaction effects that we might expect.

The current study represents the first large-scale effort focusing on brain development in youth with diagnosed CD, as opposed to conducting a dimensional analysis with smaller numbers of children exhibiting elevated conduct problems (e.g., Albaugh et al., 2023; Oostermeijer et al., 2016). Nonetheless, these cross-sectional findings should be followed up with longitudinal analyses using larger sample sizes and including younger as well as older participants. Future research could also seek to disentangle the effects of CD and ODD symptoms on brain structure and development (Hawes et al., 2023). It was not possible to investigate this issue here because most of the participants with CD had comorbid ODD (72% of the boys and 77% of the girls).

This study demonstrates that youth with CD present widespread reductions in cortical surface area (especially in temporal, frontal and parietal lobes) relative to their typically developing counterparts. Structural alterations in these regions may contribute to the impairments in emotion recognition, empathy, decision-making, and moral reasoning that are characteristic of CD and its subtypes. While the study’s primary aim was to investigate brain maturation in youth with CD compared to healthy controls, no convincing evidence was found for group differences in brain development (or three-way interactions between group, age and sex). Future research should use longitudinal research designs to investigate these issues further and broaden the developmental window to consider younger children, as we may have started assessing the children too late to detect altered brain development.