- Email: [email protected]
Medial orbital frontal lesions in late-onset depression
BRIEF REPORT Medial Orbital Frontal Lesions in Late-Onset Depression James R. MacFall, Martha E. Payne, James E. Provenzale, and K. Ranga R. Krishnan Background: Early studies using magnetic resonance (MR) imaging suggested that subcortical vascular changes are more prevalent in late-life depression and that they may play a role in the pathophysiology of depression. Studying the location of the lesion relative to the occurrence of depression could be critical in delineating the neuroanatomic substrates of depression. Our purpose was to characterize these lesions in terms of location by development of statistical parametric maps of lesions that differentiate patients from control subjects. Methods: Magnetic resonance images were acquired on 88 elderly depressed subjects (“patients,” unipolar major depression assessed using the Duke Depression Evaluation Schedule, age range 63– 80 years) enrolled in the Duke University Clinical Research Center for the Study of Late-Life Depression and 47 age- and gender-matched nondepressed subjects (“control subjects”). The MR protocol includes a volumetric, dual-contrast fast spin– echo pulse sequence. A statistical parametric map was formed from a two-group t test to test for differences in lesion density between patients and control subjects. Additional testing was performed to evaluate whether there were regions that correlated with the severity of depression using the 17-item Hamilton Depression rating. Results: The statistical parametric mapping analysis between groups showed two major regions of increased lesion density in the patients in the medial orbital prefrontal white matter. Severity of depression among depressed patients was correlated with lesions in the medial orbital region. Conclusions: This study supports recent evidence implicating the medial orbital frontal cortex in depression. Biol Psychiatry 2001;49:803– 806 © 2001 Society of Biological Psychiatry Key Words: Frontal cortex, lesions, depression, neuroanatomy, elderly, vascular
From the Departments of Radiology (JRM, JEP) and Psychiatry and Behavioral Sciences (MEP, KRRK), Duke University Medical Center, Durham, North Carolina. Address reprint requests to K. Ranga R. Krishnan, M.D., Duke University Medical Center, Department of Psychiatry and Behavioral Sciences, Box 3950, Durham NC 27710. Received August 15, 2000; revised November 20, 2000; accepted November 29, 2000.
© 2001 Society of Biological Psychiatry
Introduction
T
he causes of depression in the elderly are poorly understood (Byrum et al 1999). Genetic factors are of less significance in patients presenting with depression for the first time late in life (Krishnan et al 1997). Thus, other factors are believed to play a role in the etiopathophysiology of late-onset depression. Early studies using magnetic resonance imaging (MRI) suggested that subcortical vascular changes are more prevalent in late-life depression and that they may play a role in the pathophysiology of depression (Krishnan et al 1997). The importance of these changes has also been demonstrated in a recent epidemiologic study (Steffens et al 1999). Studying the location of the lesion relative to the occurrence of depression could be critical in delineating the neuroanatomic substrates of depression. The purpose of this study was to characterize these lesions in terms of location by development of statistical parametric maps of lesions that differentiate patients from control subjects.
Methods and Materials Magnetic resonance images were acquired for 88 elderly depressed subjects (age range 63– 80 years, 63 female), assessed using the Duke Depression Evaluation Schedule (George et al 1989) and enrolled in the Duke University Clinical Research Center for the Study of Late-Life Depression, and 47 age- and gender-matched nondepressed subjects (control subjects) after obtaining written informed consent under a protocol approved by the Duke Institutional Review Board. All patients met DSM-IV criteria for major depression. The mean age of the patients was 72.6 ⫾ 7.9 years. The mean age of the control subjects was 72.21 ⫾ 6.25 years. The mean age of onset of depression was 49.3 ⫾ 22.6 years. Sixty-two of the patients had recurrent major depression. Fifty-two had melancholia. The MR protocol includes a volumetric, dual-contrast fast spin– echo pulse sequence. Imaging was performed at 1.5 T on a commercial MRI system (GE Signa, GE Medical Systems, Milwaukee) with the standard quadrature head coil. Pulse sequence parameters were repetition time ⫽ 3000 msec; echo time ⫽ 30, 130 msec; echotrain length ⫽ 16; field of view ⫽ 22 cm; slice thickness ⫽ 3 mm; slices acquired without gaps; imaging bandwidth ⫽ ⫾16 kHz, number of excitations ⫽ 1; 256 ⫻ 256 matrix. Volumes of various brain 0006-3223/01/$20.00 PII S0006-3223(00)01113-6
804
J.R. MacFall et al
BIOL PSYCHIATRY 2001;49:803– 806
regions were identified by processing the imaging data on a SUN workstation using the MRX computer program (GE Medical Systems, modified by Kikinis et al 1992) and a semiautomated method that took advantage of the dual-echo image contrasts so that tissue classes (gray matter, white matter, cerebrospinal fluid, and lesion) were identified by operator selection. The process produces a segmented image in which each pixel is assigned to one of the tissue types or to a background value. The interrater reliability for large regions such as the total brain volume is .98, whereas for smaller regions such as the hippocampus or the lesion volume it is .86. Statistical parametric mapping (SPM) was performed on a SUN workstation using SPM99 (Wellcome Department of Cognitive Neurology, London) (Wright et al 1995). An image containing just the regions identified as containing white matter lesions was created from the segmented image. The nominal pixel intensity of a lesion region was set to an arbitrary value of 128, with nonlesion regions having a value of zero. This image was spatially normalized by applying the matrix determined from normalizing the corresponding proton density (PD) image to the SPM PD template image provided in the SPM release. The transformation distorts the lesion nominal intensity value; thus regions having nonzero intensity were restored to a nominal value of “128” after the normalization. Each normalized image was smoothed with a low-pass three-dimensional spatial filter with a kernel value of five pixels in the x, y, and z directions (Wright et al 1995). The filtered lesion images thus represent “lesion density” in a given voxel within these approximations. Within the original area the density is high, with the density falling off away from the center of the region. A statistical parametric map was formed from a two-group t test to test for differences in lesion density between depressed and control subjects. The significance level was set at p ⬍ .01. Analysis of the total lesion volume between depressed and control subjects showed the two groups to have nearly identical total lesion volumes; hence, no global effects of total lesion volume were included in the analysis. Additional testing was performed to evaluate whether there were regions that correlated with the severity of depression using the 17-item Hamilton Depression rating. The presence of lesions was also assessed using the Coffey rating scale (Krishnan et al 1997).
Results There was a higher rate of vascular disease in the control subjects (34) who were recruited to have a greater risk for vascular disease than in patients (38; 2 ⫽ 10.8, p ⬍ .001). Subcortical hyperintensity was greater in patients than in control subjects (0 and 1 vs. 2 and 3; 2 ⫽ 7.32, p ⬍ .006). Large deep white matter hyperintensities were also greater in depressed patients, although this did not reach significance (3 vs. rest, 2 ⫽ 2.85, p ⬍ .09). The SPM analysis between groups showed two major regions of increased lesion density in the patients in the medial orbital prefrontal white matter as shown in Figure 1. The right-side region has an uncorrected p ⬍ .001 and a corrected value
Figure 1. Regions for which lesion density is higher in patients than in control subjects superimposed on a typical image.
for the whole brain of p ⬍ .016, whereas the left-side region has an uncorrected p ⬍ .087 and a corrected value of p ⬍ .980. Limiting the number of voxels by considering only the prefrontal region gives corrected values of p ⬍ .007 and p ⬍ .120 for right and left sides, respectively. The results of the second analysis, performed using only patients to see if there were regions that correlated with their depression scores, are shown in Figure 2. These data indicate that even when considering the whole brain—that is, not restricting the analysis to a preselected region—the corrected p value is significant (⬍.005) for the medial orbital prefrontal region and for a left-side region in the internal capsule.
Discussion This study, in conjunction with other literature on the role of this part of the prefrontal cortex, suggests a critical role for these lesions in the pathophysiology of depression in late life. Previous studies with rating scales have provided evidence that large lesions, especially in the frontal cortex and basal ganglia, are related to depression. This study is the first to more precisely localize the lesions. The regions identified are consistent with other data in the literature (Ongur et al 1998) that indicate involvement of the medial orbital prefrontal region and the neural pathways to caudate and other limbic regions. The postmortem studies that demonstrate a reduction in thickness and reduction in size of neurons and neuronal cell bodies and glial density in the medial orbital frontal cortex (Rajkowska et al 1999) and data from our group (Lai et al 2000) showing a reduction in volume of the medial orbital frontal cortex provide additional evidence suggesting that this region of the brain may be important in the pathophysiology of depression.
Late-Onset Depression
BIOL PSYCHIATRY 2001;49:803– 806
805
Figure 2. Statistical parametric mapping (SPM99) results for lesion areas correlate with depression score.
The findings contribute to the growing evidence that the orbital cortex is critically involved in affective disorders. Previous regional cerebral blood flow and metabolic studies demonstrated increased metabolism in patients with milder forms of depression and decreased metabolism in more ill patients. Serotonin depletion produces regional cerebral blood flow changes and metabolic changes in these same regions of the brain (Bremner et al 1997). Numerous studies have suggested that the pyramidal cells of the orbital cortex play a role in extinguishing unreinforced responses to appetitive stimuli (Rolls 1995; Rolls et al 1994). This probably involves an interaction with the amygdala and other limbic structures. Neuropsychologic assessment of humans with orbital cortex lesions has demonstrated impaired performance on tasks evaluating emotional performance (Angrilli et al 1999). Subjects also exhibited perseveration and difficulty shifting intel-
lectual strategies in response to changing demands. These studies complement and are consistent with findings in animal studies (Rolls et al 1994). The combination of image segmentation, spatial normalization, and statistical parametric map creation allows the identification of probable areas where lesions may directly relate to depression. Future research will be directed toward functional assessment of these white matter regions in depression, which may lead to understanding whether lesions that have functional correlates in these regions are associated with the problem of treatmentrefractory depression.
This study was sponsored by the National Institute of Mental Health (Grant No. MH40159).
806
BIOL PSYCHIATRY 2001;49:803– 806
References Angrilli A, Palomba D, Cantagallo A, Maietti A, Stegagno L (1999): Emotional impairment after right orbitofrontal lesion in a patient without cognitive deficits. Neuroreport 10:1741– 1746. Bremner JD, Innis RB, Salomon RN, Staib LH, Ng CK, Miller HL, et al (1997): Positron emission tomography measurement of cerebral metabolic correlates of tryptophan depletioninduced depressive relapse. Arch Gen Psychiatry 54:346 – 374. Byrum CE, Ahearn EP, Krishnan KR (1999): A neuroanatomic model for depression. Prog Neuropsychopharmacol Biol Psychiatry 23:175–193. George LK, Blazer DG, Hughes DC, Fowler N (1989): Social support and the outcome of major depression. Br J Psychiatry 154:478 – 485. Kikinis R, Shenton ME, Gerig G, Martin J, Anderson M, Metcalf D, et al (1992): Routine quantitative analysis of brain and cerebrospinal fluid spaces with MR imaging. J Magn Reson Imaging 2:619 – 629. Krishnan KR, Hays JC, Blazer DG (1997): MRI-defined vascular depression. Am J Psychiatry 154:497–501. Lai T-J, Payne ME, Byrum CE, Steffens DC, Krishnan KRR
J.R. MacFall et al
(2000): Reduction of orbital frontal cortex volume in geriatric depression. Biol Psychiatry 48:971–975. Ongur D, Drevets WC, Price JL (1998): Glial reduction in the subgenual prefrontal cortex in mood disorders. Proc Natl Acad Sci U S A 95:13290 –13295. Rajkowska G, Miguel-Hidalgo JJ, Wei J, Dilley G, Pittman SD, Meltzer HY, et al (1999): Morphometric evidence for neuronal and glial prefrontal cell pathology in major depression. Biol Psychiatry 45:1085–1098. Rolls ET (1995): A theory of emotional and consciousness, and its application to understanding the neural basis of emotion. In: Gazzaniga MS, editor. The Cognitive Neurosciences. Cambridge, MA: MIT Press, 1091–1106. Rolls ET, Hornak J, Wade D, McGrath J (1994): Emotion-related learning in patients with social and emotional changes associated with frontal lobe damage. J Neurol Neurosurg Psychiatry 57:1518 –1524. Steffens DC, Helms MJ, Krishnan KR, Burke GL (1999): Cerebrovascular disease and depression symptoms in the cardiovascular health study. Stroke 30:2159 –2166. Wright IC, McGuire PK, Poline JB, Travere JM, Murray RM, Frith CD, et al (1995): A voxel-based method for the statistical analysis of gray and white matter density applied to schizophrenia. Neuroimage 2:244 –252.
From the Departments of Radiology (JRM, JEP) and Psychiatry and Behavioral Sciences (MEP, KRRK), Duke University Medical Center, Durham, North Carolina. Address reprint requests to K. Ranga R. Krishnan, M.D., Duke University Medical Center, Department of Psychiatry and Behavioral Sciences, Box 3950, Durham NC 27710. Received August 15, 2000; revised November 20, 2000; accepted November 29, 2000.
© 2001 Society of Biological Psychiatry
Introduction
T
he causes of depression in the elderly are poorly understood (Byrum et al 1999). Genetic factors are of less significance in patients presenting with depression for the first time late in life (Krishnan et al 1997). Thus, other factors are believed to play a role in the etiopathophysiology of late-onset depression. Early studies using magnetic resonance imaging (MRI) suggested that subcortical vascular changes are more prevalent in late-life depression and that they may play a role in the pathophysiology of depression (Krishnan et al 1997). The importance of these changes has also been demonstrated in a recent epidemiologic study (Steffens et al 1999). Studying the location of the lesion relative to the occurrence of depression could be critical in delineating the neuroanatomic substrates of depression. The purpose of this study was to characterize these lesions in terms of location by development of statistical parametric maps of lesions that differentiate patients from control subjects.
Methods and Materials Magnetic resonance images were acquired for 88 elderly depressed subjects (age range 63– 80 years, 63 female), assessed using the Duke Depression Evaluation Schedule (George et al 1989) and enrolled in the Duke University Clinical Research Center for the Study of Late-Life Depression, and 47 age- and gender-matched nondepressed subjects (control subjects) after obtaining written informed consent under a protocol approved by the Duke Institutional Review Board. All patients met DSM-IV criteria for major depression. The mean age of the patients was 72.6 ⫾ 7.9 years. The mean age of the control subjects was 72.21 ⫾ 6.25 years. The mean age of onset of depression was 49.3 ⫾ 22.6 years. Sixty-two of the patients had recurrent major depression. Fifty-two had melancholia. The MR protocol includes a volumetric, dual-contrast fast spin– echo pulse sequence. Imaging was performed at 1.5 T on a commercial MRI system (GE Signa, GE Medical Systems, Milwaukee) with the standard quadrature head coil. Pulse sequence parameters were repetition time ⫽ 3000 msec; echo time ⫽ 30, 130 msec; echotrain length ⫽ 16; field of view ⫽ 22 cm; slice thickness ⫽ 3 mm; slices acquired without gaps; imaging bandwidth ⫽ ⫾16 kHz, number of excitations ⫽ 1; 256 ⫻ 256 matrix. Volumes of various brain 0006-3223/01/$20.00 PII S0006-3223(00)01113-6
804
J.R. MacFall et al
BIOL PSYCHIATRY 2001;49:803– 806
regions were identified by processing the imaging data on a SUN workstation using the MRX computer program (GE Medical Systems, modified by Kikinis et al 1992) and a semiautomated method that took advantage of the dual-echo image contrasts so that tissue classes (gray matter, white matter, cerebrospinal fluid, and lesion) were identified by operator selection. The process produces a segmented image in which each pixel is assigned to one of the tissue types or to a background value. The interrater reliability for large regions such as the total brain volume is .98, whereas for smaller regions such as the hippocampus or the lesion volume it is .86. Statistical parametric mapping (SPM) was performed on a SUN workstation using SPM99 (Wellcome Department of Cognitive Neurology, London) (Wright et al 1995). An image containing just the regions identified as containing white matter lesions was created from the segmented image. The nominal pixel intensity of a lesion region was set to an arbitrary value of 128, with nonlesion regions having a value of zero. This image was spatially normalized by applying the matrix determined from normalizing the corresponding proton density (PD) image to the SPM PD template image provided in the SPM release. The transformation distorts the lesion nominal intensity value; thus regions having nonzero intensity were restored to a nominal value of “128” after the normalization. Each normalized image was smoothed with a low-pass three-dimensional spatial filter with a kernel value of five pixels in the x, y, and z directions (Wright et al 1995). The filtered lesion images thus represent “lesion density” in a given voxel within these approximations. Within the original area the density is high, with the density falling off away from the center of the region. A statistical parametric map was formed from a two-group t test to test for differences in lesion density between depressed and control subjects. The significance level was set at p ⬍ .01. Analysis of the total lesion volume between depressed and control subjects showed the two groups to have nearly identical total lesion volumes; hence, no global effects of total lesion volume were included in the analysis. Additional testing was performed to evaluate whether there were regions that correlated with the severity of depression using the 17-item Hamilton Depression rating. The presence of lesions was also assessed using the Coffey rating scale (Krishnan et al 1997).
Results There was a higher rate of vascular disease in the control subjects (34) who were recruited to have a greater risk for vascular disease than in patients (38; 2 ⫽ 10.8, p ⬍ .001). Subcortical hyperintensity was greater in patients than in control subjects (0 and 1 vs. 2 and 3; 2 ⫽ 7.32, p ⬍ .006). Large deep white matter hyperintensities were also greater in depressed patients, although this did not reach significance (3 vs. rest, 2 ⫽ 2.85, p ⬍ .09). The SPM analysis between groups showed two major regions of increased lesion density in the patients in the medial orbital prefrontal white matter as shown in Figure 1. The right-side region has an uncorrected p ⬍ .001 and a corrected value
Figure 1. Regions for which lesion density is higher in patients than in control subjects superimposed on a typical image.
for the whole brain of p ⬍ .016, whereas the left-side region has an uncorrected p ⬍ .087 and a corrected value of p ⬍ .980. Limiting the number of voxels by considering only the prefrontal region gives corrected values of p ⬍ .007 and p ⬍ .120 for right and left sides, respectively. The results of the second analysis, performed using only patients to see if there were regions that correlated with their depression scores, are shown in Figure 2. These data indicate that even when considering the whole brain—that is, not restricting the analysis to a preselected region—the corrected p value is significant (⬍.005) for the medial orbital prefrontal region and for a left-side region in the internal capsule.
Discussion This study, in conjunction with other literature on the role of this part of the prefrontal cortex, suggests a critical role for these lesions in the pathophysiology of depression in late life. Previous studies with rating scales have provided evidence that large lesions, especially in the frontal cortex and basal ganglia, are related to depression. This study is the first to more precisely localize the lesions. The regions identified are consistent with other data in the literature (Ongur et al 1998) that indicate involvement of the medial orbital prefrontal region and the neural pathways to caudate and other limbic regions. The postmortem studies that demonstrate a reduction in thickness and reduction in size of neurons and neuronal cell bodies and glial density in the medial orbital frontal cortex (Rajkowska et al 1999) and data from our group (Lai et al 2000) showing a reduction in volume of the medial orbital frontal cortex provide additional evidence suggesting that this region of the brain may be important in the pathophysiology of depression.
Late-Onset Depression
BIOL PSYCHIATRY 2001;49:803– 806
805
Figure 2. Statistical parametric mapping (SPM99) results for lesion areas correlate with depression score.
The findings contribute to the growing evidence that the orbital cortex is critically involved in affective disorders. Previous regional cerebral blood flow and metabolic studies demonstrated increased metabolism in patients with milder forms of depression and decreased metabolism in more ill patients. Serotonin depletion produces regional cerebral blood flow changes and metabolic changes in these same regions of the brain (Bremner et al 1997). Numerous studies have suggested that the pyramidal cells of the orbital cortex play a role in extinguishing unreinforced responses to appetitive stimuli (Rolls 1995; Rolls et al 1994). This probably involves an interaction with the amygdala and other limbic structures. Neuropsychologic assessment of humans with orbital cortex lesions has demonstrated impaired performance on tasks evaluating emotional performance (Angrilli et al 1999). Subjects also exhibited perseveration and difficulty shifting intel-
lectual strategies in response to changing demands. These studies complement and are consistent with findings in animal studies (Rolls et al 1994). The combination of image segmentation, spatial normalization, and statistical parametric map creation allows the identification of probable areas where lesions may directly relate to depression. Future research will be directed toward functional assessment of these white matter regions in depression, which may lead to understanding whether lesions that have functional correlates in these regions are associated with the problem of treatmentrefractory depression.
This study was sponsored by the National Institute of Mental Health (Grant No. MH40159).
806
BIOL PSYCHIATRY 2001;49:803– 806
References Angrilli A, Palomba D, Cantagallo A, Maietti A, Stegagno L (1999): Emotional impairment after right orbitofrontal lesion in a patient without cognitive deficits. Neuroreport 10:1741– 1746. Bremner JD, Innis RB, Salomon RN, Staib LH, Ng CK, Miller HL, et al (1997): Positron emission tomography measurement of cerebral metabolic correlates of tryptophan depletioninduced depressive relapse. Arch Gen Psychiatry 54:346 – 374. Byrum CE, Ahearn EP, Krishnan KR (1999): A neuroanatomic model for depression. Prog Neuropsychopharmacol Biol Psychiatry 23:175–193. George LK, Blazer DG, Hughes DC, Fowler N (1989): Social support and the outcome of major depression. Br J Psychiatry 154:478 – 485. Kikinis R, Shenton ME, Gerig G, Martin J, Anderson M, Metcalf D, et al (1992): Routine quantitative analysis of brain and cerebrospinal fluid spaces with MR imaging. J Magn Reson Imaging 2:619 – 629. Krishnan KR, Hays JC, Blazer DG (1997): MRI-defined vascular depression. Am J Psychiatry 154:497–501. Lai T-J, Payne ME, Byrum CE, Steffens DC, Krishnan KRR
J.R. MacFall et al
(2000): Reduction of orbital frontal cortex volume in geriatric depression. Biol Psychiatry 48:971–975. Ongur D, Drevets WC, Price JL (1998): Glial reduction in the subgenual prefrontal cortex in mood disorders. Proc Natl Acad Sci U S A 95:13290 –13295. Rajkowska G, Miguel-Hidalgo JJ, Wei J, Dilley G, Pittman SD, Meltzer HY, et al (1999): Morphometric evidence for neuronal and glial prefrontal cell pathology in major depression. Biol Psychiatry 45:1085–1098. Rolls ET (1995): A theory of emotional and consciousness, and its application to understanding the neural basis of emotion. In: Gazzaniga MS, editor. The Cognitive Neurosciences. Cambridge, MA: MIT Press, 1091–1106. Rolls ET, Hornak J, Wade D, McGrath J (1994): Emotion-related learning in patients with social and emotional changes associated with frontal lobe damage. J Neurol Neurosurg Psychiatry 57:1518 –1524. Steffens DC, Helms MJ, Krishnan KR, Burke GL (1999): Cerebrovascular disease and depression symptoms in the cardiovascular health study. Stroke 30:2159 –2166. Wright IC, McGuire PK, Poline JB, Travere JM, Murray RM, Frith CD, et al (1995): A voxel-based method for the statistical analysis of gray and white matter density applied to schizophrenia. Neuroimage 2:244 –252.