Anterior cingulate volume in pediatric bipolar disorder and autism

1. Introduction 

It is suggested that a reduced left anterior cingulate gyrus (ACG) volume may be related to the brain structure endophenotype (McDonald et al., 2004) of bipolar disorder (Hasler et al., 2006). However, only three studies show possible reductions in ACG volume while a number of studies demonstrate no significant differences (Riffkin et al., 2005). Using automated voxel based morphometry, researchers identified volume reductions of the dorsal lateral prefrontal gyrus (Dickstein et al., 2005), medial temporal lobe, orbitofrontal cortex, and ACG in adolescents with BD compared to age-matched healthy controls (Wilke et al., 2004). Only one study using manual tracing methods found that the left ACG was significantly smaller in pediatric BD compared to typically developing children (Kaur et al., 2005) and several others found no difference (Frazier et al., 2005a). The purpose of this study is to evaluate the validity of the left ACG volume as an endophenotype of BD by attempting to control for potential confounds not included in previous studies using a similar age autism comparison group.

The lack of significant reductions in left ACG volume in some studies may be due to the medication exposure of the subjects. Controversy exists over the effects that psychotropic medications have on brain anatomy, with studies reporting increases, decreases, or no changes in the same structure for various diagnostic groups. The neuroprotective effects of lithium may contribute to the absence of gray matter reduction in some studies of adults diagnosed with BD (Kaur et al., 2005). The effect of typical antipsychotics on grey matter volume is inconsistent across disorders, with some studies finding decreases (Dazzan et al., 2005, Lieberman et al., 2005) while others report increases. Atypical antipsychotics may affect the volume of the anterior cingulate in BP, but to our knowledge, no studies have reported this effect.

Reduced intelligence quotient (IQ) may be an additional confounding factor in determining whether differences are found in ACG volume. For example, prefrontal gray matter volume has been found to be positively correlated with IQ in children aged 4 to 18 years (Reiss et al., 1996). The most robust findings in schizophrenia, BD, and post-traumatic stress disorder continue to be global reductions in brain volume that correlate with decreased IQ (De Bellis et al., 2002, DelBello et al., 2004, Frazier et al., 2005a, Mehler and Warnke, 2002). Taken together, the literature suggests that reductions in IQ are not disease-specific, and additional factors may account for global reductions in brain volume across diagnostic groups. Studies have not considered the combined effects of IQ and medication on ACG volume in pediatric BD.

The current study evaluated ACG volume in two clinical populations, pediatric BD and autism spectrum disorder (ASD), as compared to children with no psychiatric diagnosis (NP), in order to evaluate the effects of medication, IQ, and diagnosis on ACG volumes. The ASD group was chosen as a positive control for several reasons (Towbin et al., 2005, Stahlberg et al., 2004, McCracken et al., 2002): 1) Overall functional impairment and symptom severity are similar in ASD and BD, but there is different core symptomatology; 2) both disorders occur in childhood; and 3) at least in our sample, there is a similar exposure to medications (see below). We used these two diagnostic groups along with a control group without psychiatric diagnosis to examine factors that may have contributed to the conflicting pattern of volumetric differences in the ACG previously reported in the literature. We hypothesized that children with BD will have smaller ACG volumes regardless of age, psychotropic medication history, or IQ compared with children with ASD or NP.

2. Materials and methods 

3. Results 

The BD, ASD, and NP groups were similar in mean age, gender, and intracranial volume. The BD and ASD groups were similar in mean IQ, while the mean IQ of the NP group was significantly higher than either diagnostic group (Table 1).

Table 1. Demographics for diagnostic groups

	Diagnosis	Sample size (N)	Gender⁎⁎	Age	FSIQ	Intracranial volume (cm3)
				(mean±SD)	(mean±SD)	(mean±SD)
	BD	16	M=12	10.63±4.56	100.81±12.56	1231.7±112.0
			F=4
	ASD	24	M=22	10.50±1.93	97.96±16.22	1296.9±109.0
			F=2
	NP	15	M=12	10.94±1.65	114.67±9.82⁎	1262.1±126.3
			F=3

⁎ = significant at p<0.05; control>ASD and BD; FSIQ = full-scale IQ; BD = bipolar disorder; ASD = autism spectrum disorder; NP = children with no psychiatric diagnosis. ⁎⁎ = Fisher's exact (exact sig. 2 sided) p=1.00 between NP and BD; p=0.354 between NP and ASD.

The ANCOVA compared log ACG volumes in the ASD, BD, and NP groups with intracranial volume (ICV), age, gender, and FSIQ as covariates. For the left log ACG volume, a significant difference was found between groups (p=0.001, df=2, F=7.66; covariates: ICV, p<0.001; age, p=0.285; gender, p=0.077; IQ, p=0.349). Post hoc comparison revealed that the left log ACG volume was significantly smaller in the BD group compared to both the NP (p=0.004; effect size=1.48cm3) and ASD groups (p=0.006; effect size=1.66cm3) (Table 2). No group mean differences were found in the right ACG volumes.

Table 2. Left and right anterior cingulate gyrus volumes

	Diagnosis	Sample size (N)	L_ACG	R_ACG	Significance (p-value)
			mean±SD	mean±SD	BD vs. ASD		BD vs. NP		ASD vs. NP
			(cm3)	(cm3)	L_ACG	R_ACG	L_ACG	R_ACG	L_ACG	R_ACG
	BD	16	5.23±1.09	7.17±1.89	0.004⁎	1.00	0.006⁎	1.00	1.00	1.00
	ASD	24	7.32±2.62	7.40±1.69
	NP	15	6.86±1.15	7.26±1.18

⁎ = significant at p<0.05; significance levels derived from univariate ANCOVAS performed on log-transformed ACG volume data with age, gender, FSIQ, and intracranial volume as covariates; Bonferroni adjustments performed to adjust for multiple comparisons. R_ACG = right anterior cingulate gyrus; L_ACG = left anterior cingulate gyrus; BD = bipolar disorder; ASD = autism spectrum disorder; NP = children with no psychiatric diagnosis. Data are reported as mean volumes.

As shown in Table 3, an ANCOVA with ACG volume as the dependent variable, lifetime medication usage and diagnosis (BD or ASD) as the independent variables, and ICV, age, gender, and FSIQ as covariates, revealed no significant medication effects on either the right or left ACG. Using pairwise comparisons with Bonferroni correction, the BD group exposed to atypical medications was not statistically significantly different from the NP group. The BD groups both positive and negative for lifetime mood stabilizer exposure and negative for lifetime atypical neuroleptic exposure were significantly smaller in the left ACG volume compared to the NP group (p<0.022). Overall, individuals with BD had significantly smaller mean left ACG volume irrespective of medication history when compared to individuals with ASD who were both medicated and non-medicated (Table 3).

Table 3. Means for left and right anterior cingulate gyrus volumes based upon lifetime medication usage

	Medication	Diagnosis	Ever taken?	Frequency	Mean left ACG volume (cm3)	Mean right ACG volume (cm3)
	Mood stabilizer	BD	Yes	9	5.27±1.13a	7.50±1.51
			No	5	4.83±0.70a	6.94±2.71
		ASD	Yes	4	6.80±2.47	8.29±1.43
			No	19	7.57±2.70	7.26±1.76
	Atypical neuroleptic	BD	Yes	8	5.57±1.02	7.51±1.08
			No	6	4.52±0.58a	7.03±2.15
		ASD	Yes	5	7.57±2.74	8.00±1.40
			No	18	7.40±2.67	7.28±1.80

Two BD subjects and one ASD subject were excluded from analysis due to missing medication information.

a Significantly different (p<0.022) from NP.

4. Discussion 

In this study, the left, but not the right anterior cingulate gyrus was significantly smaller in the pediatric BD group than in both NP control children and children with ASD. The differences between the BD and NP groups replicate results recently reported by three separate studies (Kaur et al., 2005, Sassi et al., 2004, Wilke et al., 2004). The results of this study extend previous findings by indicating that the left ACG volume is smaller in the BD group when compared to an age and IQ matched ASD group with similar medication exposure.

A growing literature suggests involvement of the ACG in cognitive control, one of the many processes important for attention (Carter et al., 2000, Sharp et al., 2006, Whittle et al., 2006). Cognitive control deficits interfere with the ability to sustain attention and ignore distracting stimuli (Quraishi and Frangou, 2002, Whittle et al., 2006). Changes in fMRI activation of the ACG correlate with deficits in cognitive control tasks in neuroimaging studies of BD (Egner and Hirsch, 2005, Lie et al., 2006). Furthermore, subjects with BD perform more poorly on tests of executive function than normal controls, which is related to the volume of the ACG for the BD group (Zimmerman et al., 2006).

Dickstein et al. (2005) speculated that there may be a biological correlation between changes in the left side of the brain and depression. While reduction in left subgenual cingulate volume has been found in pediatric depression (Botteron et al., 2002), no differences in subgenual volumes were reported for a pediatric BD sample (Kaur et al., 2005, Sanches et al., 2005). Kaur et al. found the left anterior cingulate to be smaller in pediatric BD while excluding the subgenual region from measurement (Kaur et al., 2005). The current study differs from these prior studies because our tracing protocol takes into account anatomic heterogeneity of the ACG by defining four variants that include the subgenual region (Gittins and Harrison, 2004, Riffkin et al., 2005).

In the study by Kaur et al., the mean volume of the BD group's left ACG without the subgenual region was 2.49 cm3 (Kaur et al., 2005). From the same group, the mean value for the BD group's left subgenual cingulate region itself was 0.28 cm3 (Sanches et al., 2005). In our study, the mean left ACG volume is 5.23 cm3, which is still much greater than these two volumes combined (2.49 cm3+0.28 cm3=2.77 cm3). The functional meaning of these differences in ACG boundaries requires further study (Fornito et al., 2004). However, it is unlikely that the subgenual region alone contributes to differences in the ACG volume observed in this study.

This study protocol for measuring ACG volume, compared to the previously mentioned studies, extends the caudal margin of the ACG past the anterior commissure (Kaur et al., 2005, Sanches et al., 2005). The paracingulate, when a double cingulate is present, is also included in the ACG volume. Prior studies have shown that the paracingulate markedly influences ACG volume and is often excluded for lack of an ideal segmentation method (Yucel et al., 2001). Future studies should look at the effect of asymmetry in the paracingulate on bipolar disorder, as asymmetry in this structure may have functional significance (Fornito et al., 2004).

A limitation to this study is that the mean IQ in the NP group was significantly higher than in the ASD or BD groups, similar to many pediatric imaging studies (Frazier et al., 2005b, Kumra et al., 2000). Brain structure volume often positively correlates with IQ (Reiss et al., 1996); differences in IQ may account for changes in brain volume. However, the difference in left ACG volume between the NP and BD groups remained after adjusting for IQ differences. More importantly, the mean IQ was similar between the ASD and BD groups and the left ACG volume remained significantly smaller in the BD group compared to the ASD group.

Another limitation to this study is that medication exposure in the ASD and BD groups was similar but not matched or adjusted for lifetime dose equivalents. Due to the sample size, length of time on medication could not be taken into account but subjects were on a stable dose for at least 4 weeks prior to assessment. In adult schizophrenia, increased typical antipsychotic use was associated with increased ACG volumes (Kopelman et al., 2005). Atypical antipsychotic use has also been reported to decrease ACG volume over time (McCormick et al., 2005). In this study, the left ACG volume in individuals with BD exposed to atypical antipsychotic medications was not significantly different from the NP group.

In summary, the left ACG volume is significantly smaller in individuals with BD than age-matched youths with ASD or no psychiatric diagnosis. These findings should be confirmed by larger studies that follow subjects over time, pre- and post-medication exposure. Reduction in left ACG volume may be associated with the increased cognitive dysfunction and mood disorder associated with BD.