Posterior orbitofrontal sulcogyral pattern associated with orbitofrontal cortex volume reduction and anxiety trait in panic disorder

Abstract

Aims:  The posterior region of the orbitofrontal cortex (OFC), which forms its sulcogyral pattern during neurodevelopment, receives multisensory inputs. The purpose of the present study was to assess the relationship between posterior OFC sulcogyral pattern and OFC volume difference in patients with panic disorder.

Methods:  The anatomical pattern of the posterior orbital sulcus (POS) was classified into three subtypes (absent POS, single POS, double POS) using 3-D high-spatial resolution magnetic resonance images obtained from 28 patients with panic disorder and 28 age- and gender-matched healthy controls. Optimized voxel-based morphometry (VBM) was performed to assess OFC volume differences between the two groups by subtype. Categorical regression analysis was applied to examine the association of POS subtypes with State–Trait Anxiety Inventory and Revised Neuroticism-Extraversion-Openness Personality Inventory scores.

Results:  No significant difference was found in POS subtype distribution between control subjects and patients with panic disorder. VBM, however, indicated volume reduction in the right posterior–medial OFC region in panic disorder patients with absent POS and single POS. Single POS was positively associated with Trait-Anxiety (β = 0.446, F = 6.409, P = 0.020), and absent POS was negatively associated with Trait-Anxiety (β = −0.394, F = 5.341, P = 0.032) and Neuroticism trait (β = −0.492, F = 6.989, P = 0.017).

Conclusions:  POS subtypes may be relevant to volume reduction in OFC and the anxiety trait in patients with panic disorder. These findings suggest that volume reduction in OFC in panic disorder may be associated with neurodevelopment.

THE ORBITOFRONTAL CORTEX (OFC) has direct reciprocal connections with the amygdala,¹ which plays an important role in neuroanatomical pathways for panic disorder.² The amygdala also has connections with a region of the anterior cingulate cortex known as the affective division.³ The OFC mediates anxiety behavior and perception in child and adolescent primates,⁴ and is involved in emotional processing in humans.⁵ Anxiety requires multiple sensory inputs, which enter the posterior region of the OFC from the primary sensory cortex.⁵ These findings suggest that anatomical differences may be related to OFC dysfunction.

The OFC sulcogyral pattern, which exhibits anatomical variability among individuals, is determined in early neurodevelopment,⁶ and develops during 16–44 weeks gestation.⁷ It has been reported that the OFC continues to mature longer than other ventral brain regions,^5,8 although frontal gray matter volume decreases after the age of 10 years.⁹ The posterior region of the OFC develops before the anterior region.^8,9 A recent volumetric study demonstrated that the OFC sulcogyral pattern was associated with brain volume in schizophrenia.¹⁰ These findings indicate a relationship between sulcogyral pattern and OFC gray matter volume. Few magnetic resonance imaging (MRI) volumetric studies of the OFC have involved patients with panic disorder, whereas several MRI volumetric studies of the amygdala^11,12 and anterior cingulated cortex¹³ have been done.

Chiavaras and Petrides evaluated OFC sulcogyral patterns of 50 healthy volunteers, and reported considerable pattern differences among individuals.¹⁴ Based on these findings, the anatomical pattern of the posterior orbital sulcus (POS) was classified as absent POS, single POS, or double POS, according to the presence/absence and number of POS in the OFC.

In the present study we hypothesized that anatomical variability may be associated with dysfunction of the OFC in patients with panic disorder. This study focused on the structure of the OFC posterior region, which receives multiple sensory inputs and modulates the anxiety response. We investigated differences in the distribution of POS subtypes between patients with panic disorder and healthy control subjects. Additionally, we used optimized voxel-based morphometry (VBM) to investigate OFC volume differences between patients and control subjects of each POS subtype group. We also assessed the association of POS subtypes with clinical measures for anxiety and personality traits.

METHODS

Subjects

Twenty-eight patients with panic disorder (male, n = 10; female, n = 18; mean age, 38.4 ± 9.8 years) were recruited from the Department of Psychiatry at Yokohama City University Hospital. Twenty-eight age- and gender-matched healthy control (HC) subjects (male, n = 10; female, n = 18; mean age, 37.8 ± 9.8 years) were also recruited. All subjects were right-handed. The panic disorder diagnosis was based on DSM-IV criteria. The Structured Clinical Interview for DSM-IV Axis I Disorders (SCID-I) was used to ascertain the diagnosis.¹⁵Table 1 lists the demographic and clinical characteristics of these groups. Comorbid psychiatric diseases included agoraphobia (n = 15), current major depression (n = 1), previous dysthymia (n = 1), and previous major depression (n = 6). Twenty-five patients were taking antidepressants and benzodiazepine. HC subjects were recruited from the community and the staff of Yokohama City University and were screened using the SCID non-patient edition¹⁶ and Mini-International Neuropsychiatric Interview.¹⁷ HC subjects had neither previous nor present psychiatric diseases.

Table 1.  Subject characteristics

Variables	Patients with panic disorder (n = 28)		Healthy control subjects (n = 28)		t-test or χ^2	d.f.	P
	Mean	SD	Mean	SD
*P < 0.05, **P < 0.01.^†t-test;^‡χ^2 test. Handedness >0 if right-handed, as assessed on the Edinburgh Inventory. NEO PI-R, Revised NEO Personality Inventory; PDSS, Panic Disorder Severity Scale; STAI A Trait, Anxiety trait of the State–Trait Anxiety Inventory; Total IQ, total intelligence quotient, as assessed on the Wechsler Adult Intelligence Scale–Revised.
Age (years)	38.4	9.8	37.8	9.8	0.23^†	54	0.82
Gender (male/female)	10/18		10/18		0.00^‡	1	1.00
Handedness	1.0	0.1	1.0	0.1	0.24^†	54	0.81
Subject socioeconomic status	2.43	0.879	2.00	0.980	1.694^†	52	0.096
Parents' socioeconomic status	2.57	0.879	2.23	0.710	1.559^†	52	0.125
Total IQ	103.7	13.8	109.9	13.4	−1.59^†	46	0.12
STAI A Trait	46.5	10.9	35.4	6.4	4.52^†	51	<0.01**
NEO PI-R Neuroticism	110.2	26.1	88.0	22.8	3.22^†	48	<0.01**
NEO PI-R Extraversion	101.7	18.1	109.0	15.6	−1.53^†	48	0.13
NEO PI-R Openness to Experience	114.7	17.0	113.6	14.3	0.26^†	48	0.80
NEO PI-R Agreeableness	110.8	15.6	112.6	13.6	−0.43^†	48	0.67
NEO PI-R Conscientiousness	106.5	16.3	110.1	15.6	−0.78^†	48	0.44
PDSS	9.0	4.9
Antidepressant dosage (amitriptyline equivalent, mg)	74.0	66.3
Anti-anxiety agent dosage (diazepam equivalent, mg)	6.60	4.66

As part of a comprehensive neuropsychological battery, the total IQ scores of subjects from both groups (patients, n = 24; HC, n = 24) were determined using the Wechsler Adult Intelligence Scale–Revised.¹⁸ Anxiety and personality traits were assessed using the Trait-Anxiety subscale of State–Trait Anxiety Inventory (STAI; patients, n = 28; HC, n = 27)¹⁹ and the Revised Neuroticism-Extraversion-Openness Personality Inventory (NEO; patients, n = 24; HC, n = 26).²⁰ We also used the Panic Disorder Severity Scale (PDSS)²¹ as a clinical measure. The present study was approved by the Yokohama City University Review Board. Written informed consent was obtained from all subjects in advance of study participation.

Magnetic resonance imaging

MRI was done with a 1.5-T Siemens Magnetom Symphony scanner (Siemens Medical System, Erlangen, Germany) at Yokohama City University Hospital. The 128 contiguous sagittal T1-weighted images were acquired using turbo-FLASH sequences with the following parameters: echo time = 3.93 ms, repetition time = 1960 ms, inversion time = 1100 ms, flip angle = 15°, field of view = 24 cm, matrix = 256 × 256 × 128, and voxel dimension = 0.9375 × 0.9375 × 1.5 mm.

Sulcogyral pattern identification

Sulcogyral pattern identification was based on a previous report by Chiavaras and Petrides.¹⁴ To clarify the OFC, identification of several sulci was required, including the medial orbital sulcus (MOS), lateral orbital sulcus (LOS), transverse orbital sulcus (TOS), olfactory sulcus (Olf), intermediate orbital sulcus (IOS), POS, and sulcus fragmentosus (Fr). POS was often not detected (i.e. absent POS). When a single POS was observed (i.e. single POS), it emerged from the posterior orbital convexity, or from the sequence of TOS, the caudal portion of MOS, and the caudal portion of LOS. When two POS were observed (i.e. double POS), both POS emerged from the posterior orbital convexity or one POS emerged from the sequence of TOS, the caudal portion of MOS, and the caudal portion of LOS, and another emerged from the posterior orbital convexity (Fig. 1). Intraclass correlation coefficients (Cronbach's α) for subtype classification (absent, single, and double POS) were 0.942 for the left OFC and 0.960 for the right OFC based on three raters (T.R., A.F., and R.S.) blinded to diagnosis.

Figure 1. The presence and number of posterior orbital sulci (POS) are defined by three subtypes: absent POS, single POS, and double POS. (a) Dotted red line shows the location of the POS in the orbitofrontal cortex (OFC). (b) Sulci of the OFC include the olfactory sulcus (Olf), medial orbital sulcus (MOS; r, rostral; c, caudal), POS, transverse orbital sulcus (TOS), lateral orbital sulcus (LOS), intermediate orbital sulcus (IOS; m, medial; l, lateral), and sulcus fragmentosus (Fr). (c) Absent POS; (d,e) single POS; (f,g) double POS; (h) absent POS. Magnetic resonance imaging of (i) single POS, and (j) double POS in the left OFC. Purple line, POS; Blue line, MOSc, TOS and LOSc. (k) The control group had the same POS subtype distribution. In addition, no significant group difference was observed between patients with panic disorder and healthy control subjects. (a–g adapted and modified from a previous paper by Chiavaras and Petrides.¹⁴)

Statistical analysis

Demographic data and POS subtypes

To assess differences between patients and controls, t-tests compared demographical data including age, total IQ, handedness, Anxiety (A) Trait score, and NEO score. χ² test was used to assess group differences in gender frequencies. As shown in Fig. 1, using the Chiavaras and Petrides results¹⁴ as the expected number for observed distributions of POS subtypes, we applied the χ² test to evaluate differences in POS subtype distributions between subjects in the Chiavaras and Petrides study¹⁴ and the present control subjects. We also used the χ² test to assess the difference in POS subtype distributions between the present patients and control subjects (Fig. 1).

Optimized voxel-based morphometry

Optimized VBM was used to determine OFC gray matter volume differences between patients and control subjects. Previous whole brain VBM carried out at Department of Psychiatry, Yokohama City University School of Medicine had already detected significant gray matter volume reduction in the right ventromedial prefrontal cortex, which includes the OFC, in patients with panic disorder compared with healthy subjects.²² In the present study we focused on the OFC, and group differences were assessed separately within each POS subtype group because the previous structural MRI study suggested an association between sulcogyral pattern and brain volume.¹⁰

Optimized VBM was carried out using Statistical Parametric Mapping (SPM) 2 (Wellcome Department of Cognitive Neurology, London, UK). Briefly, a study-specific template was created by averaging segmented images of all subjects in the POS groups. Next, all T1-weighted images were normalized onto this template non-linearly and then segmented into gray matter. After non-brain voxels were removed from the gray matter images, these images were modulated and smoothed using a 12-mm full-width at half-maximum Gaussian kernel.

Group effects were assessed using analysis of covariance (ANCOVA). Intracranial volume was treated as a confounding covariate to correct for global anatomical variations. Age and gender, which are believed to influence brain structure,^22,23 were also treated as covariates. The general linear model²⁴ was used and, based on our previous findings,²² VBM was performed for the OFC using WFU Pickatlas software.²⁵ Given the small number of subjects in each POS subtype group, uncorrected P < 0.001 was considered statistically significant to determine OFC gray matter volume differences between patients with panic disorder and control subjects within each POS subtype group.

Correlation

Categorical regression analysis was used to determine the extent to which absent POS, single POS, and double POS (nominal variables) predicted anxiety and personality trait test scores in patients and control subjects, and clinical symptoms and clinical measurements in patients. These three nominal variables were entered as independent variables in a single model of categorical regression with each measure entered as a dependent variable within each study group. To reduce the risk of false positives, contributions of all three nominal variables to variance in each dependent variable were tested in a single model of categorical regression, rather than multiple univariate comparisons.

RESULTS

POS subtypes

The POS subtype distribution in 28 control subjects did not differ from the distribution of POS subtypes reported by Chiavaras and Petrides (total of left and right, χ² = 0.443, P = 0.801; left, χ² = 0.253, P = 0.881; right, χ² = 0.281, P = 0.869; Fig. 1).¹⁴ Comparing patients to control subjects in the present study, no significant group difference was detected in the POS distribution of total, left, and right OFC (total, χ² = 0.000, P = 1.000; left, χ² = 2.908, P = 0.234; right, χ² = 4.000, P = 0.135).

Voxel-based morphometry

In the absent-POS group (patient : control = 9:10), a gray matter volume reduction in the right medial–posterior OFC (Brodmann area [BA] 11: peak coordinates [x, y, z (mm)] = 2, 26, −12; peak Z-score = 3.26, no. voxels = 24) was detected in patients compared with control subjects (Fig. 2). In the single-POS group (patient : control = 24:26), patients had smaller gray matter volumes in the right medial–posterior OFC (BA11: peak coordinates [x, y, z (mm)] = 6, 24, −10; peak Z-score = 3.95, no. voxels = 20; BA47: peak coordinates [x, y, z (mm)] = 16, 20, −12; peak Z-score = 3.25; no. voxels = 10). In the double-POS group (patient : control = 7:7), no group difference in OFC volume was observed. No OFC regions increased in volume in patients compared with control subjects in any subtype group.

Figure 2. (a,b) Group comparison of the absent posterior orbital sulcus (POS) group on voxel-based morphometry (VBM) for orbitofrontal cortex (OFC). Both (a) sagittal and (b) axial views showed gray matter volume reduction of the right medial–posterior OFC in patients with panic disorder compared with healthy control subjects (peak coordinates [x,y,z (mm) = 2, 26, –12]. (c–f) Group comparisons in the single-POS group. (c,e) Sagittal view; (d,f) axial view. Gray matter volume reductions were detected in the right medial–posterior OFC in patients with panic disorder compared with healthy control subjects. (c,d) Peak coordinates (x,y,z [mm], 6, 24, –10, Brodmann area [BA]11); (e,f) peak coordinates (x,y,z [mm], 16, 20, –12, BA47).

The subjects in each of the three groups (i.e. absent, single, double POS) had either the same POS subtype in both hemispheres of the OFC (i.e. homo-subtype) or a different POS subtype in each hemisphere (i.e. hetero-subtype), therefore the total number in the figures was unmatched with the number of subjects.

Categorical regression

Patients with panic disorder obtained significantly higher A Trait scores (t = 4.52, P < 0.01) and NEO neuroticism trait scores (t = 3.22, P < 0.01) compared with control subjects (Table 2). single POS was positively associated with STAI A Trait (β = 0.446, F = 7.174, P = 0.014), and absent POS was negatively associated with A Trait (β = −0.437, F = 7.214, P = 0.014) and NEO Neuroticism trait scores (β = −0.520, F = 7.450, P = 0.013) only in patients with panic disorder.

Table 2.  Categorical regression analysis

Dependent Variables	Patients with panic disorder					Healthy control subjects
	ANOVA	Independent variables	β	F	P	ANOVA	Independent variables	β	F	P
* P < 0.05, ** P < 0.01. NEO PI-R, Revised NEO Personality Inventory; PDSS, Panic Disorder Severity Scale; POS, posterior orbital sulcus; STAI A Trait, Anxiety trait of the State–Trait Anxiety Inventory.
STAI A Trait	F(3,22) = 9.48 P < 0.01	Absent POS	−0.437	7.214	0.014*	F(3,23) = 3.88 P = 0.022	Absent POS	0.378	3.826	0.063
		Single POS	0.446	7.174	0.014*		Single POS	−0.265	2.303	0.143
		Double POS	−0.437	1.320	0.263		Double POS	−0.223	1.307	0.265
NEO PI-R Neuroticism	F(3,20) = 5.97 P < 0.01	Absent POS	−0.520	7.450	0.013*	F(3,22) = 2.42 P = 0.094	Absent POS	0.314	2.165	0.155
		Single POS	0.259	1.678	0.210		Single POS	−0.283	2.234	0.149
		Double POS	−0.092	0.260	0.615		Double POS	−0.153	0.504	0.485
NEO PI-R Extraversion	F(3,20) = 2.56 P = 0.084	Absent POS	0.099	0.196	0.663	F(3,22) = 3.62 P = 0.029	Absent POS	−0.623	9.560	<0.01**
		Single POS	−0.470	4.026	0.059		Single POS	−0.227	1.618	0.217
		Double POS	0.017	0.007	0.936		Double POS	0.189	0.865	0.363
NEO PI-R Openness to Experience	F(3,20) = 5.09 P < 0.01	Absent POS	−0.752	10.89	<0.01**	F(3,22) = 7.29 P < 0.01	Absent POS	0.417	5.719	0.026*
		Single POS	−0.269	0.056	0.209		Single POS	−0.550	12.63	<0.01**
		Double POS	−0.098	0.010	0.605		Double POS	−0.051	0.083	0.776
NEO PI-R Agreeableness	F(3,20) = 7.57 P < 0.01	Absent POS	0.824	18.47	<0.01**	F(3,22) = 2.86 P = 0.060	Absent POS	−0.199	0.906	0.352
		Single POS	0.577	10.70	<0.01**		Single POS	0.142	0.584	0.453
		Double POS	0.121	0.022	0.481		Double POS	−0.605	8.252	0.009
NEO PI-R Conscientiousness	F(3,20) = 2.95 P = 0.058	Absent POS	0.428	5.362	0.164	F(3,22) = 5.09 P < 0.01	Absent POS	−0.058	0.095	0.761
		Single POS	0.367	3.524	0.089		Single POS	0.267	2.538	0.125
		Double POS	0.540	9.442	0.009		Double POS	0.652	11.66	<0.01**
Agoraphobia	F(3,24) = 2.26 P = 0.107	Absent POS	−0.089	0.186	0.670
		Single POS	0.434	4.372	0.047
		Double POS	0.124	0.393	0.536
PDSS	F(3,23) = 1.43 P = 0.260	Absent POS	−0.428	3.469	0.075
		Single POS	0.268	1.395	0.250
		Double POS	0.061	0.094	0.762
Antidepressant dosage (amitriptyline equivalent, mg)	F(3,24) = 1.05 P = 0.388	Absent POS	−0.058	0.070	0.793
		Single POS	0.107	0.234	0.633
		Double POS	0.323	2.360	0.138
Anti-anxiety agent dosage (diazepam equivalent, mg)	F(3,24) = 1.26 P = 0.311	Absent POS	0.158	0.528	0.474
		Single POS	0.182	0.690	0.414
		Double POS	0.402	3.741	0.065

Absent POS was negatively associated with the NEO openness-to-experience trait (β = −0.752, F = 10.89, P < 0.01) in patients with panic disorder. The NEO agreeableness trait was positively associated with absent POS (β = 0.824, F = 18.47, P < 0.01) and single POS (β = 0.577, F = 10.70, P < 0.01) in patients with panic disorder.

In control subjects, absent POS was negatively associated with the NEO extraversion trait (β = −0.623, F = 9.560, P < 0.01). absent POS was positively associated with the NEO openness-to-experience trait (β = 0.417, F = 5.719, P = 0.026), while single POS was negatively associated (β = −0.550, F = 12.63, P < 0.01). Double POS was positively associated with the NEO conscientiousness trait (β = 0.652, F = 11.66, P < 0.01).

POS subtypes were not associated with agoraphobia, PDSS, or the dosage of antidepressant and anti-anxiety agents.

DISCUSSION

In the present study we demonstrated that POS subtype is associated with posterior OFC volume reduction and anxiety in patients with panic disorder, although POS subtype distribution did not differ between the panic disorder and the control subjects. There have been inconsistent results for POS subtype distribution in panic disorder. No difference in POS subtype distribution was found in several studies.^26,27 These studies suggested that acquired causes such as traumatic life events were also associated with the development of panic disorder symptoms. Studies in twins, however, have reported inherent different vulnerabilities to the same environmental factor among patients.^28,29 OFC has been reported to be involved with guidance of reward-related behavior, expectation of reinforcements for goal-directed behavior, and decision making and motivation. These functions are processed primarily in the anterior OFC and require multisensory inputs from the posterior OFC.⁵ In addition, the central nucleus of the amygdala receives multisensory inputs and acts as the autonomic mediator for the brainstem nuclei related to panic attack.³⁰ Our previous study showed that volume reduction in the right central nucleus may be associated with dysfunction in patients with panic disorder.¹² Because the OFC has been reported to be interconnected with the amygdala,¹ the OFC and amygdala may work together to mediate anxiety. The present study found significant gray matter volume reduction in the right posterior OFC in the absent-POS and single-POS groups of panic disorder patients, suggesting right posterior OFC dysfunction.

The OFC is connected to the anterior cingulate cortex, which is also connected to the amygdala.³ Our previous study indicated gray matter volume reduction of the right dorsal region of the anterior cingulate cortex, which is associated with cognitive function, whereas no reduction was found in the rostral ventral region, which is associated with emotion and affective processing.¹³ In the present study, however, VBM demonstrated that absent-POS and single-POS groups exhibited gray matter volume reduction in the right medial–posterior region of the OFC, which plays a role in the independent motivational system.⁵ This region is also located near the affective division of the anterior cingulate cortex. In the present study we found that single POS was positively associated with A Trait, and absent POS was negatively associated with A Trait. VBM also showed that single POS had more gray matter volume reduction than absent POS in OFC. A recent large cohort study found that premature infants had gray matter volume reduction.³¹ Additionally, adolescents with very low birthweight had increased anxiety.³² These previous findings may be relevant to the present findings. Given the anxiety-mediating role of the posterior OFC, these findings suggest that immaturity of the POS may be associated with anxiety in patients with panic disorder.

The present study has several limitations. Because comorbid depression may affect the anatomical and clinical features of panic disorder, it is necessary to investigate differences between patients with panic disorder and without major depression/dysthymia. Large studies are needed to perform VBM using statistically rigorous threshold Family-Wise Error (FWE)-correction.

In conclusion, we have demonstrated that the posterior OFC may play an important role in input of multisensory information to mediate anxiety. POS subtypes in the posterior OFC are associated with OFC volume reduction and anxiety in patients with panic disorder, but POS subtype distribution did not differ between patients and control subjects. These findings suggest that OFC volume reduction may be associated with neurodevelopment in patients with panic disorder.

ACKNOWLEDGMENTS

The present study was supported by a grant-in-aid from the Ministry of Health, Labor and Welfare of Japan awarded to Y.H. and in part by a Grant-in-Aid for Scientific Research (No. 14370294) from the Ministry of Education, Science, Sports and Culture, Japan.