SEARCH

SEARCH BY CITATION

Keywords:

  • voxel-based morphometry;
  • migraine;
  • chronic migraine;
  • cerebral pain network

Abstract

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. COMMENTS
  6. Acknowledgments
  7. REFERENCES

Background.— Migraine is generally considered a functional brain disorder lacking structural abnormalities. Recent magnetic resonance imaging (MRI) studies, however, suggested that migraine may be associated with subtle brain lesions.

Objective.— We evaluated the presence of global or focal gray or white matter alterations in migraine patients using voxel-based morphometry (VBM), a fully automated method of analyzing changes in brain structure. VBM data also were used to evaluate possible differences between episodic and chronic migraine.

Methods.— Twenty-seven migraine right-handed patients and 27 healthy controls were selected for the study. Sixteen patients fulfilled the International Headache Society criteria for episodic migraine and 11 for chronic migraine. MRI scans were analyzed with MATLAB 6.5 and SPM2 software, using VBM method.

Results.— In comparison with controls, migraineurs presented a significant focal gray matter reduction in the Right Superior Temporal Gyrus, Right Inferior Frontal Gyrus, and Left Precentral Gyrus. Chronic migraine patients, compared to episodic, showed a focal gray matter decrease in the bilateral Anterior Cingulate Cortex, Left Amygdala, Left Parietal Operculum, Left Middle and Inferior Frontal Gyrus, Right Inferior Frontal Gyrus, and bilateral Insula. Considering all the migraine patients, a significant correlation between gray matter reduction in anterior cingulate cortex and frequency of migraine attacks was found.

Conclusions.— Our study shows that migraine is associated with a significant gray matter reduction in several of the cortical areas involved in pain circuitry. In addition, we found a significant correlation between frequency of migraine attacks and signal alteration in the Anterior Cingulate Cortex. Our data provide new insight into migraine pathophysiology and support the concept that migraine may be a progressive disorder.

Abbreviations:
MRI

magnetic resonance imaging

VBM

voxel-based morphometry

WHO

World Health Organization

ICHD-II

Second Edition of International Classification of Headache Disorders

MPRAGE

Magnetization Prepared Rapid Gradient Echo

GM

gray matter

WM

white matter

CSF

cerebrospinal fluid

ANCOVA

analysis of covariance

TIV

total intracranial volume

MNI

Montreal Neurological Institute

BA

Brodman area

ACC

Anterior Cingulate Cortex

PAG

periaqueductal gray matter

Migraine is a primary headache disorder characterized by recurrent attacks of throbbing pain associated with neurological, gastrointestinal, and autonomic symptoms.1 In the U.S.A., migraine affects approximately 18% of females and 6% of males.2 According to the World Health Organization (WHO), migraine ranks among the world's most disabling medical illnesses.3 Migraine is generally considered an episodic brain disorder. However, approximately 20% of migraine patients develop chronic migraine, a condition characterized by the presence of frequent headache attacks.4 Patients with chronic migraine demonstrate remarkable impairment of their daily activities and are severely burdened by their headache syndrome.5

According to the Second Edition of International Classification of Headache Disorders (ICHDII)6 structural brain lesions are absent in primary headaches. Contrarywise, recent studies with voxel-based morphometry (VBM), a fully automated method of analyzing changes in brain structure, demonstrated selective brain alterations in both cluster headache and chronic tension-type headache. Cluster headache is characterized by a bilateral increase of gray matter in posterolateral hypothalamus7 while, in patients with chronic tension-type headache, a gray matter decrease was found in orbifrontal cortex, insula, and anterior cingulate cortex.8 In patients with migraine a previous VBM study did not detect any change in the structure of the brain.9 However, several functional imaging studies have shown that, during migraine attacks, there is an abnormal metabolism in several cortical and subcortical brain regions, suggesting a dysfunctional pain system.10,11

The purpose of this study was to investigate the presence of structural brain abnormalities in patients with migraine using the optimized VBM method, a highly sensitive technique to detect focal gray and white matter abnormalities in the brain.12,13 We compared the VBM data of a group of right-handed migraine patients, recruited from an university-based headache center, with those of healthy controls and we searched for global and local differences in gray and white matter. We further analyzed VBM data of the patients in order to look for structural brain differences in migraine subgroups and we evaluated correlations with the clinical characteristics of the disease.

METHODS

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. COMMENTS
  6. Acknowledgments
  7. REFERENCES

Patients.— Twenty-seven right-handed migraine patients (21 females, 6 males; mean age ± SD = 34.9 ± 8.4 yrs) were selected for the study. Twenty-one patients fulfilled the ICHD-II criteria for migraine without aura and 6 for migraine with aura. Mean age at onset of the disease was 14.2 ± 4.6 yrs, mean frequency of migraine attacks was 11.8 ± 9.7 days per month, mean duration of migraine attacks was 27.3 ± 15.5 hours, and mean duration of the disease was 20.6 ± 8.9 yrs. For additional analyses, migraine patients were divided into 2 subgroups, based on the monthly attack rate (more or less than 15 days per month): A. episodic migraine: 16 patients (12 females, 4 males; mean age ± SD = 32.1 ± 8.7 yrs); B. chronic migraine: 11 patients (9 females, 2 males; mean age ± SD = 38.9 ± 6.4 yrs). The patients with chronic migraine were on prophylactic medication. No patient had drug abuse and no patient had macroscopic brain T2-visible lesions on MRI scans. A group of 27 right-handed healthy subjects, age, and sex matched (20 females, 7 males; mean age ± SD = 34.9 ± 8.6 yrs) served as control. The controls were screened by a neurologist specialized in headaches in order to exclude abnormal neurological findings, migraine and other primary headaches. The protocol of this study was reviewed and approved by the Medical Ethics Committee of the San Giovanni Battista Hospital of Torino and written informed consent was obtained from all participants.

MRI Scans and Voxel-Based Morphometry.— Magnetic resonance imaging was performed with a 1 tesla Siemens MAGNETOM IMPACT scanner. A 3-D structural MRI was acquired on each subject using a T-1 weighted Magnetization Prepared Rapid Gradient Echo (MPRAGE) sequence14 with following parameters: TR = 11.4 ms, TE = 4.4, TI = 300 ms, flip angle = 15°, matrix = 256 × 256, 128 sagittal slices, slice thickness = 1.4 mm, in plane resolution = 1 mm × 1 mm. The same scanner and the same scanning protocol were used for all participants. Data were analyzed using MATLAB 6.5 (the MathWorks Inc., Sherborn, USA; http://www.mathworks.com) and SPM2 (Wellcome Department of Cognitive Neurology, London; http://www.fil.ion.ucl.ac.uk/spm). Optimized Voxel-Based Morphometry method is characterized by several mathematical analytical steps (data preprocessing), described in details by previous studies:14,15 first, anatomical and gray/white matter customized templates were created using structural scans of both healthy controls and migraine patients; after, optimal spatial normalization parameters were estimated16 using customized templates and applied to structural brain images; segmentation into gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF) compartments17 incorporating brain extraction of normalized brain images was performed; an additional correction for volume changes (modulation) was performed on segmented images in order to preserve the local tissue morphology; finally, segmented brain images were smoothed with an isotropic Gaussian kernel of 12 mm full width at half maximum. Additionally, global GM and WM volumes and total intracranial volume (TIV) of each subject were estimated using segmented images in native space.

Statistical Analysis.— Student t test was used to compare global GM and WM volumes between migraine patients and controls and between migraine subgroups. The segmented gray/white matter images were analyzed using statistical parametric mapping (SPM2) employing the framework of the General Linear Model.18 Analysis of covariance (ANCOVA) was used to compare migraine patients with the control group and migraine subgroups. TIV for modulated images and global mean voxel value of segmented images for unmodulated images were respectively used as nuisance variables in statistical comparisons, in order to detect only regionally specific changes in gray and white matter. Moreover, a voxel by voxel regression analysis was performed, to delineate a possible relationship between local gray matter changes and clinical data of overall migraine patients group. As chronic migraine patients are significantly older than episodic migraine patients (t = 2.23 and P = .035), age was added as nuisance variable in statistical comparison between these groups and in statistical correlation with clinical data. Volumetric changes were tested by incorporating the modulation of segmented data. An absolute voxel signal intensity threshold masking of 0.2 was set up. Significance threshold, corrected for multiple comparisons using Family Wise Error method, was set at P < .05. Based on our prior hypothesis regarding the involvement of pain transmitting structures, the analysis was repeated with a significance threshold set at P < .001, without correction for multiple comparisons, throughout the whole brain, applying a small volume correction (S.V.C.) for multiple comparisons in these areas, with a threshold of P < .05. For each design matrix, significant effects in the negative and positive directions were considered.

Localization of Statistical Parametric Maps.— Significant clusters obtained from SPM2 were neuro-anatomically localized into stereotactic Montreal Neurological Institute (MNI) space using Talairach Daemon Client version 2.0 (http://ric.uthscsa.edu/projects/talairachdaemon.html) with MATLAB function (available in http://www.mrccbu.cam.ac.uk/Imaging/Common/mnispace.shtml). Neuro-anatomical localizations were visually checked comparing statistical parametric maps (overlaid by structural MRI images normalized into stereotactic MNI space) with Talairach atlas.19

RESULTS

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. COMMENTS
  6. Acknowledgments
  7. REFERENCES

Structural MRI scans were normal in all the patients and the controls and no artifact was found. There was no significant difference in global GM and WM volumes between cases (mean volume ± SD: 657 ± 62 and 446 ± 62 mL, respectively) and controls (mean volume ± SD: 683 ± 69 and 452 ± 38 mL, respectively). Comparing global volumes between migraine subgroups, we found only a significant GM reduction in chronic migraine in comparison with episodic migraine (mean volume ± SD: 628 ± 53 and 678 ± 60 mL, respectively; t = −2.23 and P = .035). This result can be related to the mean age difference between these migraine subgroups. We detected no significant difference in regional white matter in all comparisons performed with SPM2.

Local Gray Matter Comparison Between Migraine Patients and Controls.—Table 1 shows regions of significant gray matter reduction, identified by MNI coordinates, in migraine patients in comparison with healthy controls using modulated data. Significant clusters were found in the Right Superior Temporal Gyrus with extension to the Parietal Operculum (Brodman area – BA – 42 and 43), in the Right Inferior Frontal Gyrus (BA 44 and 45), and in the Left Precentral Gyrus (BA 44). Figure 1 shows these differences identified by Statistical Parametric Map. No clusters of significant gray matter increase were found using modulated data. No clusters of significant gray matter difference were identified using unmodulated data.

Table 1.—. Regions of Significant Gray Matter Reduction in Migraine Patients Versus Controls Using Modulated Images (T > 3.26 with 51 Degrees of Freedom, P < .05 with S.V.C.)
CECoordinates (mm)T51Anatomic Region (Local Maxima)
xyz
  1. The cluster extension (CE), representing the number of contiguous voxels passing the height threshold, was set at ≥10. Coordinates in bold represent a cluster with the peak T-value within the cluster. Subsequent nonbold coordinates identify further peaks within the same cluster that meet the significance level. Brain regions are indicated by Montreal Neurological Institute coordinates.

270967−654.46Right Superior Temporal Gyrus, Area 42
65−644.38Right Superior Temporal Gyrus, Area 42
65−674.37Right Transverse Temporal Gyrus, Area 42
66−7104.30Right Parietal Operculum, Area 43
64−9183.74Right Parietal Operculum, Area 43
64−9203.73Right Parietal Operculum, Area 43
6136314153.97Right Inferior Frontal Gyrus, Area 44
6315193.86Right Inferior Frontal Gyrus, Area 45
6216113.75Right Inferior Frontal Gyrus, Area 44
581483.37Right Inferior Frontal Gyrus, Area 44
10−6011123.38Left Precentral Gyrus, Area 44
image

Figure 1.—. Differences in gray matter between migraine patients and healthy controls using modulated images. The background is structural T1-weighted MRI. The color coding represents T values and describes reduced gray matter. y and z represent spatial coordinates.

Download figure to PowerPoint

Local Gray Matter Comparison Between Episodic and Chronic Migraine Patients.—Table 2 shows regions of significant gray matter reduction, identified by MNI coordinates, in chronic migraine patients in comparison with episodic migraine patients using unmodulated data. The main significant cluster (1094 voxels) was found in bilateral Anterior Cingulate Cortex (ACC) (BA 32 and 24). Other smaller clusters were found in the Left Amygdala, in the Left Parietal Operculum (BA 43), in the Left Middle and Inferior Frontal Gyrus (BA 9, 10 and 44), in the Right Inferior Frontal Gyrus (BA 44), and in the bilateral Insula Lobe. No clusters of significant gray matter increase were found using unmodulated data. No clusters of significant gray matter difference were identified using modulated data.

Table 2.—. Regions of Significant Gray Matter Reduction in Chronic Versus Episodic Migraine Patients Using Unmodulated Images (T > 3.48 with 23 Degrees of Freedom, P < .05 with S.V.C.)
CECoordinates (mm)T51Anatomic region (local maxima)
xyz
  1. The cluster extension (CE), representing the number of contiguous voxels passing the height threshold, was set at ≥20. Coordinates in bold represent a cluster with the peak T-value within the cluster. Subsequent nonbold coordinates identify further peaks within the same cluster that meet the significance level. Brain regions are indicated by Montreal Neurological Institute coordinates.

1094−630154.52Left Anterior Cingulate, Area 32
 626184.10Right Anterior Cingulate, Area 24
192−21−3−244.13Left Amygdala
117−62−18184.50Left Parietal Operculum, Area 43
92−2730293.98Left Middle Frontal Gyrus, Area 9
755415344.17Right Inferior Frontal Gyrus, Area 44
6840−15−14.16Right Insula Lobe
49−5614134.03Left Inferior Frontal Gyrus, Area 44
48−44−36183.92Left Insula Lobe
28−445083.92Left Middle Frontal Gyrus, Area 10

Local Gray Matter Correlation with Clinical Data of Overall Migraine Patients Group.— The correlation between the frequency of migraine attacks and the reduction in gray matter using unmodulated data is show in Table 3. Figure 2 shows the same data, identified by the Statistical Parametric Map. A main significant cluster (863 voxels) was found in bilateral ACC (BA 24). Smaller clusters were found in the Left Amygdala, in the Left Parietal Operculum (BA 43), in the Left Middle and Inferior Frontal Gyrus (BA 9 and 44), in the Right Superior Temporal Gyrus (BA 22), in the Right Temporal Pole (BA 38), and in the bilateral Insula Lobe. No clusters of significant gray matter increase were found using unmodulated data. No clusters of significant gray matter correlation were identified using modulated data. No additional significant correlation was found with the remaining clinical features examined.

Table 3.—. Regions of Significant Correlation Between Gray Matter Reduction and the Frequency of Headache Attacks Using Unmodulated Images of Migraine Population (T > 3.48 with 23 Degrees of Freedom, P < .05 with S.V.C.)
CECoordinates (mm)T51Anatomic region (local maxima)
xyz
  1. The cluster extension (CE), representing the number of contiguous voxels passing the height threshold, was set at ≥20. Coordinates in bold represent a cluster with the peak T-value within the cluster. Subsequent nonbold coordinates identify further peaks within the same cluster that meet the significance level. Brain regions are indicated by Montreal Neurological Institute coordinates.

863−329174.31Left Anterior Cingulate, Area 24
 526174.21Right Anterior Cingulate, Area 24
320−2735305.03Left Middle Frontal Gyrus, Area 9
305−62−19175.00Left Parietal Operculum, Area 43
206−22−3−224.04Left Amygdala
91−5614134.32Left Inferior Frontal Gyrus, Area 44
6767−49183.85Right Superior Temporal Gyrus, Area 22
5541923.89Right Insula Lobe
474622−313.78Right Temporal Pole, Area 38
22−361073.92Left Insula Lobe
image

Figure 2.—. Correlation between gray matter reduction and frequency of headache attacks using unmodulated images. The background is structural T1-weighted MRI. The color coding represents T values and describes reduced gray matter. x, y, and z represent spatial coordinates.

Download figure to PowerPoint

COMMENTS

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. COMMENTS
  6. Acknowledgments
  7. REFERENCES

Our study shows that in patients with migraine there is a significant gray matter decrease in several of the brain areas that belong to the central pain network. In comparison with controls, our right-handed migraineurs presented a significant gray matter reduction in the Right Superior Temporal Gyrus with extension to the Parietal Operculum, Right Inferior Frontal Gyrus, and Left Precentral Gyrus. When we divided our patients into episodic and chronic, we found that chronic migraineurs present a gray matter decrease mainly in the bilateralACC. Other clusters were found in Left Amygdala, Left Parietal Operculum, Left Middle and Inferior Frontal Gyrus, Right Inferior Frontal Gyrus, and bilateral Insula. Finally, examining the clinical features of our patients, we found a significant correlation between frequency of migraine attacks and signal alteration in the Anterior Cingulate Cortex.

Our data are not in accord with a previous study9 showing no structural changes in the brain of patients with migraine. There are some methodological differences that may explain the discrepancies between the 2 studies. The previous study examined only patients with episodic migraine and the VBM data have been analyzed using SPM99, an older version of the statistical parametric mapping software.

Recently, Rocca et al studied a group of migraine patients showing T2-weighted MRI white matter abnormalities. Using diffusion tensor magnetic resonance imaging, a new technique with the potential to disclose subtle abnormalities in the brain, a significant reduction of gray matter density was found.20 In a following study, in agreement with our findings, optimized VBM analysis localized in the frontal and temporal lobes the areas of reduced gray matter density.21 Finally, it is of interest to note that Welch et al, using high-resolution magnetic resonance imaging, demonstrated a significant correlation between the frequency of migraine attacks and iron deposition in the periaqueductal gray matter (PAG), one of the most important centers of the descending antinociceptive neuronal network.22 These data show some similarities with the results of our study and suggest that repeated migraine attacks may induce structural abnormalities in the pain modulating structures within the central nervous system.

Modern neuroimaging techniques have produced a significant advance in our knowledge about the neural circuitry involved in nociceptive processing within the brain. There are several cortical and subcortical brain regions that are differently activated by pain in different experimental conditions, including frontal and prefrontal cortices, operculo-insular cortex, primary and secondary somatosensory cortices, anterior cingulate cortex, thalamus, insula, basal ganglia, cerebellum, amygdala, hippocampus, and regions within the parietal and temporal cortices.23,24 In our study we found a gray matter reduction in several of the centers of this cerebral pain network. Our data may contribute to explain the abnormal processing of pain as well as the reduced pain threshold found in several neurophysiological studies of patients with migraine.25-28

The anterior cingulate cortex regulates a wide variety of autonomic functions and is vital to cognitive functions, such as reward anticipation, decisionmaking, empathy, and emotion.29 Recent neuroimaging studies suggested that ACC plays a key role in the affective and attentive concomitants of pain sensations.30 The selective alteration of the cerebral structures that modulate the affective components of pain found in migraine suggests a possible neurobiological mechanism explaining the link between chronic migraine and psychiatric disturbances.31–33 In addition, our data support the results of previous studies suggesting that migraine may be considered a progressive brain disorder34,35 and suggest a radical reappraisal of the guidelines for the prophylactic antimigraine therapy, in order to avoid the progression of the disease.

The possible mechanisms underlying the gray matter reduction in migraine are currently unknown. Voxel-based morphometry detects variations of gray matter concentration per voxel as well as changes of the classification of individual voxels, eg, from white to gray matter.12,13 The observed gray matter decrease may reflect tissue shrinkage (changes in extracellular space and microvascular volume) as well as more complex processes as neurodegeneration. So, there are several possible explanations for the abnormalities observed in our patients. Gray matter changes might result from repeated ischemia caused by the cerebral blood flow abnormalities observed both during migraine attacks and in the interictal phase.36,37 On the contrary, the gray matter reduction may be a consequence of migraine specific neurotoxic mechanisms. It has been hypothesized that migraine is associated with a state of neuronal hyperexcitability, involving overactivity of the excitatory amino acids glutamate and aspartate.38 A low brain magnesium and consequent reduced gating of glutamatergic receptors could be another possible link between migraine and the mechanisms of glutamate toxicity.39 Additional longitudinal functional and structural MRI studies are needed to elucidate the mechanisms underlying gray matter abnormalities in migraine and to evaluate the possible specific physiopathologic role of structural changes found in gray matter.

In conclusion, our work indicates that the brain of migraine patients is characterized by focal structural abnormalities. VBM analysis showed that migraineurs present a significant gray matter reduction of several brain areas belonging to the pain transmitting network. In addition, we found a significant relationship between signal reduction in the anterior cingulated cortex and the frequency of headache attacks, suggesting an important role for this area in the process of migraine transformation from an episodic to a chronic brain disorder. Our data provide new insight into our understanding of migraine pathophysiology and support the concept that migraine is not just an episodic disorder but may be a chronic progressive disorder. Ongoing research and new emerging therapeutic strategies should consider this change in the conceptual model of the disease.

Acknowledgments

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. COMMENTS
  6. Acknowledgments
  7. REFERENCES

Acknowledgments: The study was supported by grants from the Ministero dell'Università e della Ricerca Scientifica and the Regione Piemonte (Italy).

The authors wish to thank all patients and healthy controls for the participation in this study, and Professor Gianni Boris Bradac for the scientific support.

Conflict of Interest: None

REFERENCES

  1. Top of page
  2. Abstract
  3. METHODS
  4. RESULTS
  5. COMMENTS
  6. Acknowledgments
  7. REFERENCES